Computer network isolator for a peripheral accessory and methods of use

The network isolator addresses the security vulnerabilities of unverified peripheral devices by enforcing strict communication policies, ensuring only authorized traffic is forwarded, thereby securing large-scale networks from compromised devices and reducing operational costs.

WO2026234537A1PCT designated stage Publication Date: 2026-11-12KLEIN BRADLEY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/IB2026/054375
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-05-04
Filing Date
2026-05-04
Publication Date
2026-11-12

AI Technical Summary

Technical Problem

Conventional network firewalls and packet-filtering gateways are ineffective in securing large-scale computer networks from compromised peripheral devices, particularly those of unknown or adversarial origin, as they do not enforce device identity or validate communication patterns, leading to potential network breaches and operational inefficiencies.

Method used

A network isolator with physically separated input/output ports and a controller that enforces strict communication policies using an allow-list database, ensuring only authorized traffic is forwarded, and unauthorized signals are blocked, even for devices of unverified origin.

Benefits of technology

The network isolator provides robust security by isolating potentially compromised peripherals, preventing unauthorized communication and data exfiltration, while allowing the use of unvetted devices without compromising network security, thus enhancing operational scalability and reducing costs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure IB2026054375_12112026_PF_FP_ABST
    Figure IB2026054375_12112026_PF_FP_ABST
Patent Text Reader

Abstract

A network isolator and method for improving computer network security are described for use with peripheral accessories whose hardware or firmware integrity cannot be fully verified. The network isolator is interposed between a single directly-attached peripheral accessory and a trusted computer network via physically separated first and second input / output ports. The isolator discovers and binds the peripheral's MAC and IP address, enforces a single-device invariant on the peripheral-side subnet, and periodically validates port isolation at runtime. All traffic traversing the isolation boundary is evaluated against an internal allow-list database specifying permitted addresses, ports, protocols, and time windows; non-matching traffic is blocked and logged. Permitted traffic may be sanitized by removing unauthorized embedded content before forwarding. A non-forwarding learning mode captures peripheral communication patterns to enable informed policy authoring.
Need to check novelty before this filing date? Find Prior Art

Description

Docket Number: BK-13-PCT COMPUTER NETWORK ISOLATOR FOR A PERIPHERAL ACCESSORY AND METHODS OF USE BACKGROUND

[0001] Without limiting the scope of the invention, its background is described in connection with computer networks, including connected peripheral accessories. More particularly, the invention describes a device and method for isolating a signal received from the peripheral accessory to safeguard the computer network from malicious intrusions and attacks.

[0002] In large-scale computer networks, the integration of peripheral devices such as video cameras, remote sensors, and similar network-connected hardware introduces significant security vulnerabilities that can be exploited by malicious actors. These devices, often categorized under the Internet of Things (loT), are typically equipped with embedded operating systems and network interfaces, making them viable entry points for unauthorized access and cyberattacks. A primary concern is that many of these peripherals operate with minimal built-in security, often using outdated firmware, hardcoded credentials, or unsecured communication protocols, which can be exploited to gain footholds within the broader network infrastructure.

[0003] Video surveillance systems, for instance, when compromised, can be used not only for unauthorized monitoring but also as relay points for advanced persistent threats (APTs). Attackers can exploit open ports or unpatched vulnerabilities in camera firmware to inject malware, perform lateral movement across internal systems, or intercept and manipulate video streams. Similarly, remote sensors — deployed for environmental monitoring, automation, or telemetry — can be compromised to exfiltrate data, introduce false readings, or serve as covert channels for command-and-control (C2) operations. These devices, often deployed at scale and in distributed locations, increase the overall attack surface and reduce visibility for network administrators, particularly if they are not integrated into centralized monitoring or endpoint protection systems.

[0004] Moreover, once compromised, such peripherals can function as persistent threats, capable of maintaining access even after detection efforts on traditional endpoints. Their role as low-privilege yet network-accessible nodes makes them attractive for reconnaissance, data leakage, and launching further attacks within the network. The growing complexity and ubiquity of these devices demand robust security architectures that include segmentation, zero-trust principles, encrypted communication, and regular firmware auditing to mitigate the risks posed by peripheralbased intrusions in enterprise-scale environments.

[0005] In large-scale and high-security environments, such as military, defense, and critical infrastructure organizations, the use of foreign-produced peripheral accessories, including network-connected video cameras, remote sensors, and other embedded devices, presents aheightened and complex threat landscape. Beyond the risks associated with software vulnerabilities and poor configuration, these devices may contain intentionally embedded malicious hardware or firmware components designed to facilitate long-term surveillance, sabotage, or unauthorized data exfiltration. This threat is particularly acute when devices are manufactured or supplied by entities operating under jurisdictions with adversarial interests or insufficient oversight.

[0006] Such foreign-manufactured peripherals may include “hardware trojans”, covertly embedded during production or supply chain transit, which can remain dormant until activated via specific network conditions, commands, or environmental triggers. These hardware implants can silently siphon data, establish covert communication channels with external threat actors, or degrade the integrity and availability of network services. In sensitive enterprise environments, such as government agencies, defense contractors, and military installations, these vulnerabilities represent not just a cyber risk, but a potential vector for “state-sponsored espionage”, sabotage, or strategic disruption.

[0007] The risk extends significantly with “surveillance devices” such as network-based security cameras, drones, and biometric access systems, which inherently gather high volumes of sensitive operational data. If compromised, they provide adversaries with direct visual or situational intelligence, such as monitoring troop movements, facility layouts, or access control protocols in real time. Similarly, remote sensors deployed in battlefield or critical infrastructure contexts, such as power grids, command centers, or nuclear facilities, can be manipulated to falsify telemetry data, disrupt automated controls, or cloak the presence of malicious activity elsewhere in the network.

[0008] Furthermore, these devices often operate outside the scope of traditional endpoint protection platforms and may be deployed in environments with limited physical or network segmentation, compounding the risk of lateral propagation following initial compromise. Attackers exploiting these devices can escalate privileges, establish long-term persistence, and evade detection by blending into normal network behavior, particularly when the malicious components are integrated at the silicon level or within proprietary firmware.

[0009] To address this multifaceted threat, high-value targets usually adopt a “zero-trust hardware policy”, emphasizing rigorous supply chain risk assessments, secure procurement practices, and detailed hardware and firmware vetting, often in collaboration with national cybersecurity authorities or trusted industry partners. Techniques such as side-channel analysis, firmware validation, and hardware attestation should be employed to verify device integrity before deployment. In addition, strict network segmentation, continuous monitoring of peripheral behavior, and prompt decommissioning of unverified foreign devices are critical to defendingagainst the strategic threats posed by potentially compromised foreign-produced peripherals in sensitive environments.

[0010] While rigorous verification of peripheral accessories and reliance on a limited pool of approved vendors are essential components of a secure hardware supply chain, especially in high-value or mission-critical environments, they introduce significant practical and operational limitations, primarily driven by cost, scalability, and technological diversity.

[0011] Cost-related constraints are among the most prohibitive. Comprehensive hardware verification, which may include techniques such as reverse engineering, side-channel analysis, firmware auditing, electromagnetic emissions testing, and secure boot validation, requires specialized facilities, skilled personnel, and highly sensitive equipment. These processes are not only time-consuming but also resource-intensive, making it economically unfeasible to apply them uniformly across all deployed devices in large-scale environments. For organizations managing thousands or millions of peripheral units ranging from sensors to imaging systems, performing deep hardware assurance on each component can be cost-prohibitive, especially when lifecycle turnover or technological upgrades occur frequently.

[0012] Additionally, reliance on a narrow pool of approved vendors, while reducing supply chain exposure, can lead to monopolistic pricing, procurement bottlenecks, and reduced innovation. Vendors with security clearances or government certifications may charge premium rates, and their production capacity may not meet the scale or specificity of all organizational needs. This constraint can also stifle access to cutting-edge or niche technologies, especially in fast-evolving sectors like Al-driven sensors or advanced imaging systems, where non-traditional or international manufacturers may lead in innovation but fall outside the approved vendor list due to geopolitical or regulatory concerns.

[0013] Moreover, vendor restrictions and costly verification processes can impair operational agility, particularly in scenarios requiring rapid deployment or replacement of hardware, such as emergency response, battlefield conditions, or dynamic industrial settings. Delays in approval cycles and sourcing challenges can hinder mission readiness or disrupt continuity of operations.

[0014] There is also a risk of over-reliance on a small set of “trusted” suppliers, which can itself become a vulnerability. A single point of compromise in a trusted supply chain, whether through insider threats, coercion, or unnoticed tampering, could have wide-reaching consequences, affecting entire fleets of devices under the assumption of security.

[0015] Conventional network firewalls and packet-filtering gateways are designed to protect a trusted internal network from threats originating on external networks, such as the public internet. These devices operate at network perimeters or between predefined network segments, evaluating packets from unknown external sources against access-control rules and permitting or blockingthem accordingly. However, conventional firewalls are not designed to address the threat posed by a single directly-attached device of unknown or adversarial provenance — a device whose hardware may itself be the source of the threat, and which is physically connected to one of the firewall's own network interfaces. In such a configuration, a conventional firewall provides no means to enforce that only one specific device is present on a given interface, to bind and validate the identity of that device at the MAC layer, to re-validate device identity when the physical link is interrupted and restored, to operate in a mode that captures the device's intended communication patterns without forwarding any of them, or to sanitize the content of permitted traffic by removing unauthorized embedded commands or covert signals before they cross the security boundary. The present invention addresses this distinct and unsolved problem by providing a network isolator specifically architected to contain the threat of a single directly- attached, potentially compromised peripheral accessory, rather than to filter traffic at a network perimeter. A further distinction from conventional firewalls is the multiplexed use of the WAN-side port. In a conventional firewall, the WAN IP address is the firewall's own management and routing address; it does not transparently represent any directly-attached device to the wider network. In the present invention, the single second input / output port (WAN port) is multiplexed to serve two simultaneous and independent purposes: (i) as the management interface of the network isolator itself, and (ii) as the transparent pass-through forwarding address for the peripheral accessory attached to the first input / output port. This multiplexing is invisible to both the peripheral accessory and the computer network, and requires no dedicated management port or separate management network.

[0016] The increasing integration of conventional peripheral devices, such as video surveillance cameras and environmental sensors, into large-scale and high-security networks creates a pressing need for novel systems and methods that can securely accommodate these widely available technologies while maintaining a high level of cybersecurity and reducing the risk of networkbased attacks.SUMMARY

[0017] Accordingly, it is an object of the present invention to overcome these and other drawbacks of the prior art by providing a novel network isolator capable of disrupting all unauthorized signals and commands that may be generated using a peripheral accessory from reaching various computers on the computer network.

[0018] It is another object of the present invention to provide a method of improving security of a computer network by providing a network isolator for attachment of one or more peripheral accessories thereto, so that the signal generated by each peripheral accessory can only be passed to the computer designated on the computer network for receiving such a signal.

[0019] It is a further object of the present invention to provide an inexpensive network isolator configured to allow the use of unvetted peripheral accessories, including foreign-made accessories, without compromising the security of the computer network itself.

[0020] In embodiments, one method for improving computer network security may include the following steps:a. providing a network isolator with a first input / output port which may be operatively connected to a peripheral accessory. The network isolator may further include a second input / output port, which may be operatively connected to the computer network. Importantly, the first input / output port is physically separated from the second input / output port to prevent unauthorized communication that may originate from the peripheral accessory. As used herein, "physical separation" between the first and second input / output ports does not require an air gap, galvanic isolation, or independent hardware enclosures. The separation may be achieved at any level of the hardware and firmware stack — from fully independent network interface silicon through to a single-chip implementation with firmware-enforced independent protocol stacks — provided that no packet or signal may traverse from the peripheral-accessory side to the computer network side without passing through, and being authorized by, the network isolator's controller and its associated traffic inspection and filtering policies; and b. operating the network isolator to evaluate all traffic arriving on the first input / output port from the peripheral accessory and all traffic arriving on the second input / output port from the computer network against an internal allow- list database, and forwarding any packet in either direction that matches a permitted entry in the allow-list database while blocking and logging all other packets, wherein the peripheral accessory may initiate communication independently and the permit or block decision depends solely on the packet's attributes relative to the allow-list database, not on whether the communication was solicited by either side.

[0021] According to the present invention, the network isolator may be configured to block all traffic, in either direction, whose source address, destination address, port, protocol, or other attributes do not match a permitted entry in the allow-list database maintained internally by the network isolator. The allow-list database may be configured to restrict traffic from the peripheral accessory to one or more designated destination computers, addresses, or address ranges on the computer network, such that traffic directed to non-permitted destinations is blocked regardless of its content or origin.

[0022] The network isolator may be also configured to inspect the signal generated by theperipheral accessory and remove unauthorized data and commands therefrom before passing the signal to the computer on the computer network designated to receive that signal.

[0023] In embodiments, one or more peripheral accessories, such as a plurality of video cameras, may be operatively attached to the network isolator via wireless or wired connections using the first input / output port. The second input / output port of the network isolator may also include one or more wired or wireless connectors or ports to facilitate its secure communication with one or more computers on the computer network. The invention is applicable to any network-addressable device whose communications with a trusted network must be strictly controlled. The term "peripheral accessory" is therefore used herein to encompass any such network-capable device, including a general-purpose computer, a server, a network router or switch, an industrial controller, or any other device requiring controlled network access, which may be deployed on the first input / output port 110 of the network isolator 100 and treated as a peripheral accessory within the scope of this invention. This flexibility allows the network isolator 100 to be used wherever a strict, policy-enforced boundary must exist between a potentially untrusted or at-risk network device and a trusted network, regardless of the device category.

[0024] In a further aspect, the second input / output port of the network isolator is multiplexed to simultaneously serve as both the management port of the isolator and the pass-through port for the peripheral accessory. Unlike a conventional firewall — in which the WAN-side address functions solely as the firewall's own management or routing address — the second input / output port of the network isolator is operatively configured to assign each arriving packet to either a management function or a pass-through forwarding function, using a traffic classifier operable on one or more of: destination transport-layer port number, destination IP address, or any combination thereof. In one method, the reserved administrative port is used as the discriminator. In another method, two IP addresses are assigned to the second input / output port, one for management and one for pass-through forwarding. Additional multiplexing methods using other packet attributes are also contemplated.BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Subject matter is particularly pointed out and distinctly claimed in the concluding portion of the specification. The foregoing and other features of the present disclosure will become more fully apparent from the following description and appended claims, taken in conjunction with the accompanying drawings. Understanding that these drawings depict only several embodiments in accordance with the disclosure and are, therefore, not to be considered limiting of its scope, the disclosure will be described with additional specificity and detail through use of the accompanying drawings, in which:

[0026] FIGURE 1 is a block diagram of the network isolator for attachment of a single peripheralaccessory to a computer network.

[0027] FIGURE 2 is a block diagram of an exemplary configuration for the attachment of several video cameras to one or more computers on a computer network.

[0028] FIGURE 3 is another exemplary configuration for the attachment of several video cameras to one or more computers on a computer network.DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS OF THE INVENTION

[0029] The following description sets forth various examples along with specific details to provide a thorough understanding of the claimed subject matter. It will be understood by those skilled in the art, however, that claimed subject matter may be practiced without one or more of the specific details disclosed herein. Further, in some circumstances, well-known methods, procedures, systems, components, and / or circuits have not been described in detail in order to avoid unnecessarily obscuring the claimed subject matter. In the following detailed description, reference is made to the accompanying drawings, which form a part hereof. In the drawings, similar symbols typically identify similar components, unless context dictates otherwise. The illustrative embodiments described in the detailed description, drawings, and claims are not meant to be limiting. Other embodiments may be utilized, and other changes may be made, without departing from the spirit or scope of the subject matter presented here. It will be readily understood that the aspects of the present disclosure, as generally described herein, and illustrated in the figures, can be arranged, substituted, combined, and designed in a wide variety of different configurations, all of which are explicitly contemplated and make part of this disclosure.

[0030] To enable the secure use of commercially available video cameras and sensors, especially in high-value, sensitive, or large-scale environments such as government installations, industrial systems, and critical infrastructure, security architectures must evolve beyond traditional trust-based or perimeter-centric approaches. The conventional model of relying exclusively on a small set of vetted vendors or performing exhaustive verification of each device is not only cost-prohibitive but also hinders operational scalability and technological innovation. In contrast, the present invention embraces a device-agnostic and behavior-driven model that prioritizes systemic resilience and dynamic threat containment over static assumptions of trust based on device origin or certification.

[0031] A cornerstone of the present invention is the implementation of a hardware-based isolation environment by using a network isolator 100 (see Figs. 1-3), which allows untrusted or semi-trusted devices, such as off-the-shelf video cameras and environmental sensors 50, to function safely while sending their signals to a secure computer network. The peripheral accessories 50 described below can be used together with the network isolator 100 to construct “sandboxed” zones that tightly control the execution context and system-level permissions of peripheral devices. In these sandboxes,peripheral accessories 50 are allowed to perform their intended functions, such as capturing video, detecting motion, or measuring temperature, while being physically isolated and programmatically restricted from accessing or influencing critical network components, sensitive data stores, or system management interfaces. As used herein, the term "peripheral accessory" is to be construed broadly and encompasses any device capable of network communication that an operator wishes to isolate from a trusted computer network by interposing the network isolator 100 between that device and the network. Peripheral accessories 50 within the scope of the invention include, without limitation: video cameras and imaging devices; environmental sensors and telemetry units; access control readers; audio capture devices; drones and unmanned vehicle control units; industrial programmable logic controllers (PLCs) and supervisory control and data acquisition (SC AD A) edge nodes; printers and multi-function peripherals; point-of-sale terminals; medical monitoring equipment; general-purpose computers and workstations that are to be quarantined or monitored; servers or server clusters that must communicate with a trusted network only through a controlled interface; network routers, switches, and access points whose management traffic is to be inspected and restricted; and any other network-addressable device whose trustworthiness cannot be fully verified at the hardware or firmware level, or whose communications with the broader network must be strictly governed by policy. The common characteristic across all peripheral accessories 50 is not their device category but rather the security relationship: they reside on the untrusted side of the network isolator 100, and all traffic between them and the trusted computer network is subject to the whitelist-based inspection, filtering, and routing policies enforced by the network isolator 100. It is therefore contemplated that any network-connected device, whether a single loT sensor or a multi-homed computer system operating a full general-purpose operating system, may serve as a peripheral accessory 50 within the meaning of the claims.

[0032] This form of containment is particularly valuable in mitigating the risks associated with supply chain attacks, firmware-level backdoors, or malicious hardware implants, which may be embedded in foreign-manufactured peripherals. Even if a peripheral accessory 50 is compromised, isolation by the network isolator 100 ensures that any malicious code is restricted to a minimal execution environment and cannot conduct lateral movement across the network, escalate privileges, or initiate unauthorized data exfiltration. This approach reflects a defense-in-depth philosophy, wherein no peripheral accessory is implicitly trusted and every interaction is mediated and monitored.

[0033] Complementing hardware-level isolation, behavioral analysis and real-time anomaly detection systems may be added to provide a dynamic method for verifying the ongoing integrity and legitimacy of device activity. Using machine learning and Al-driven models, these systems establish baseline behavior profiles for cameras and sensors, defining what "normal" looks like in terms of data rates, control signals, communication patterns, and operational outputs. Any deviation from thisnorm, such as an unexpected firmware update, unusual network communication to an external IP address, or anomalous data payload, can be flagged for investigation or trigger automated containment responses. This behavioral approach allows organizations to detect zero-day threats, firmware exploits, and covert data leaks in real time, even in devices that have not been explicitly verified at the hardware level.

[0034] The network isolator 100 is a key component of this solution, which acts as an intermediary between peripheral devices and the core computer network. One or more of the network isolators 100 may be configured to be responsible for enforcing protocol compliance, sanitizing data inputs, and verifying the authenticity of communications. By functioning as policy enforcement points, these network isolators decouple the untrusted edge (where sensors and cameras reside) from the trusted core infrastructure.

[0035] The basic configuration of the system is illustrated in Fig. 1 and represents a secure architecture for integrating one peripheral accessory (for example, a video camera or a sensor device) into a computer network while maintaining strong cybersecurity protections. Central to the system is a video filter isolator, which functions as a network isolator positioned between potentially untrusted edge devices, such as video cameras or environmental sensors, and the core computing infrastructure. While the example focuses on video cameras, the design and operation are applicable to a wide range of peripheral accessories and sensors.

[0036] The video filter isolator 100 may feature two distinct input / output ports: a first input / output port 110, which interfaces with the external peripheral device 50, and a second input / output port 150 that connects to the internal computer network infrastructure, such as a video server in this example. The first input / output port 110 may enable connection to the video camera 50 using various wired or wireless protocols, such as Ethernet 1, WiFi 1, or Bluetooth 1. The second input / output port 150 may link the video filter isolator 100 to core network components such as a video server 10, or other suitable computers and computer systems, including network- attached storage (NAS), user terminals, or network switches, as the invention is not limited in this regard. Various wired and wireless connection methods may be used for this communication, such as Ethernet 2, Virtual WiFi 2, or Virtual Bluetooth 2 protocols. In embodiments, the connection between the network isolator 100 and the peripheral accessory 50 may be different from the connection of the network isolator 100 to the computer on the computer network so that the peripheral accessory 50 may only communicate with the network isolator and not with any other computers on the computer network. Using physically separate interfaces and connections designed to completely isolate communications from the peripheral accessory ensures there is no direct pathway between the untrusted peripherals and the internal network, effectively enforcing strict network isolation.

[0037] Figures 2 and 3 illustrate alternative configurations of the network isolator system 100 inwhich multiple peripheral video cameras 50 are securely integrated into a network environment while maintaining strict isolation between the untrusted edge devices and the trusted core infrastructure. In these configurations, a network isolator 100 is used to mediate all communications between the peripheral video cameras 50 and the rest of the network. The video cameras 50 are operatively attached to the first set of input / output ports 110 of the network isolator 100, which are dedicated exclusively to peripheral device connections. These ports may accommodate multiple video cameras simultaneously, enabling the system to scale across a broad surveillance deployment or multi-sensor environment.

[0038] The input / output ports 110 of the network isolator 100 are designed as flexible, modular interfaces capable of supporting a wide range of physical and virtual connectivity types to accommodate diverse peripheral accessories. These ports may be implemented as virtual ports configured to handle various wireless communication protocols, such as WiFi, Bluetooth, or other radio-frequency-based technologies. In these configurations, the ports operate through software-defined radio modules or dedicated wireless interfaces that establish isolated, secure wireless channels to communicate with peripheral devices like video cameras, environmental sensors, or remote monitoring units.

[0039] Alternatively, input / output ports 110 may be configured as physical first connectors compatible with standard wired communication interfaces. These include commonly used cable types such as Ethernet (e.g., RJ45 connectors), USB, or serial connectors, which are designed to establish a reliable and direct connection with peripheral accessories using structured cabling. In the case of Ethernet, the ports may support Power over Ethernet (PoE), allowing a single cable to provide both data connectivity and power to the accessory, simplifying deployment in a field or industrial environments.

[0040] The dual capability of first ports 110 to operate as either virtual wireless interfaces or physical wired connectors provides the system with high adaptability and scalability. It enables secure integration of a broad array of commercial devices without requiring specialized or proprietary cabling. Regardless of the connection type, each port remains logically and electrically isolated from the core network ports, ensuring that any data received from the peripheral accessory 50 is fully inspected, filtered, and verified before being forwarded to the trusted network infrastructure. This design supports both centralized and distributed sensor deployments while maintaining strong security postures against unauthorized access or signal injection from external devices.

[0041] The second input / output ports 150 of the network isolator 100 may be architected to provide secure, controlled communication pathways between the network isolator 100 and the trusted components of a computer network. These ports may be implemented as a plurality of individual second input / output ports, with each individual port specifically configured for operative connectionto a designated computer or endpoint within the network. This one-to-one or one-to-few mapping allows the network isolator to enforce precise data routing policies and ensure that video streams or sensor outputs from individual peripheral devices are delivered only to their authorized recipients.

[0042] The second input / output ports of the network isolator may be realized in two primary forms: virtual ports and physical second connectors. In the virtual implementation, one or more second input / output ports may be configured as software-defined or virtual communication interfaces, capable of establishing secure wireless connections with corresponding computers on the network. These virtual input / output ports may operate over standard wireless protocols such as WiFi, Bluetooth, or other radio-based communication methods and are closely integrated with the system's security policies to ensure that only pre-approved, authenticated devices can access the communication channel.

[0043] Importantly, these virtual wireless input / output ports, such as Virtual Wireless First Input / Output Port or Ports and Virtual Wireless Second Input / Output Port or Ports, may be implemented using a single physical radio module. This means that a single WiFi or Bluetooth radio chip within the network isolator can support multiple virtual input / output ports simultaneously. The logical separation between these virtual ports may be achieved not by physical hardware, but by distinct encryption domains, each defined by a unique encryption key or password. By assigning different encryption credentials (for example, WPA2 or WPA3 passphrases) to each virtual first or second input / output port, the system effectively creates isolated communication channels that operate independently, even though they share the same underlying hardware and radio frequency spectrum.

[0044] This design allows the network isolator to reduce hardware complexity and cost while still enforcing strong logical and cryptographic separation between different communication paths. For example, Virtual Wireless First Input / Output Port may be used to interface with a peripheral accessory such as a video camera, while Virtual Wireless Second Input / Output Port securely transmits the filtered and verified data to a designated computer on the secure network. Though both virtual ports operate over the same frequency band and radio hardware, the encryption separation ensures that data cannot leak or be intercepted across domains.

[0045] Such an architecture not only supports efficient resource utilization but also enhances security and scalability, allowing the network isolator to serve multiple roles and maintain compartmentalization without requiring multiple discrete wireless modules. This approach is particularly advantageous in compact or embedded systems where hardware space is limited, yet high security and flexible connectivity are essential.

[0046] Alternatively, one or more second input / output ports may be configured as physical connectors, such as RJ45 Ethernet jacks, USB interfaces, Modbus serial interfaces, or other hardware-based links. These provide a dedicated, wired pathway to a specific computer or networkdevice, ensuring robust and stable connectivity while further enhancing security through physical segmentation. In some configurations, these physical ports may support additional security features such as MAC address filtering, port locking, or directional data flow enforcement to prevent any unauthorized return traffic from reaching the isolator or connected peripherals.

[0047] By offering both virtual and physical second input / output ports, the network isolator 100 may support a broad range of deployment scenarios, from wireless, decentralized systems to hardwired, high-security infrastructures, while maintaining strict isolation between untrusted peripherals and trusted network components.

[0048] Crucially, each first input / output port 110 may be physically separated from the second input / output ports 150 that connect to the internal computer network. This physical separation may involve the use of distinct network interfaces, separate internal circuitry, or air-gapped internal pathways within the network isolator, thereby preventing any direct physical, electrical, or logical bridge between the camera-facing and network-facing interfaces. This isolation strategy eliminates any potential attack vector whereby a compromised peripheral accessory could directly inject malicious traffic into the computer network. As used herein, the phrase "physically separated" with respect to the first input / output port 110 and the second input / output port 150 is to be construed broadly to encompass all architectures in which no direct, unmediated forwarding path exists between the first and second ports — regardless of whether those ports share a common physical substrate, circuit board, silicon die, or radio module. The defining characteristic of physical separation within the meaning of this invention is not the presence of an air gap, separate power rail, or independent housing, but rather the absence of any path by which a signal, packet, or frame originating on the peripheral-accessory side can reach the computer network side without first passing through, and being subject to, the inspection, filtering, authentication, and routing policies enforced by the controller of the network isolator 100. Physical separation within the meaning of this disclosure therefore encompasses, without limitation: (a) full electrical and galvanic isolation, achieved for example by optical couplers, magnetic isolation transformers, or air-gapped separate hardware modules, in which the two ports share no electrical connection whatsoever; (b) separate silicon implementations, in which the two ports are served by independent network interface controllers connected to a common processor via independent bus paths, with no shared frame-buffer memory between them; (c) a single-processor implementation, in which both ports are connected to the same microcontroller or system-on-chip but are served by independent MAC controllers, independent DMA channels, and independent memory-mapped register spaces, with the processor firmware enforcing that no packet is transferred from one port's receive buffer to the other port's transmit buffer without passing through the whitelist enforcement and filtering logic; (d) a virtualized or software-defined implementation, in which a single physical network interface or radio module supports bothports as logically distinct virtual interfaces, separated by distinct cryptographic domains, encryption keys, or network credentials such that a device on one virtual interface has no means of communicating with a device on the other virtual interface except through the controlled mediation of the network isolator's controller; and (e) any intermediate architecture combining elements of the foregoing, such as two ports on a common switch fabric or network processor in which hardware access-control lists and firmware-enforced forwarding rules prevent any direct port-to-port traffic while permitting traffic that transits the isolator's inspection and policy engine. It is expressly contemplated that in all of the foregoing embodiments the two ports may be physically co-located on the same printed circuit board, housed within the same enclosure, and connected to the same processor, provided that the sole communication path between them runs through the controller's mediated, policy-governed forwarding logic. What is excluded from the meaning of "physically separated" is any configuration in which a packet or signal may pass from the first port to the second port — or vice versa — without the knowledge, inspection, or authorization of the network isolator's controller.

[0049] Within the network isolator 100, a dedicated controller governs all communications. The controller is programmed to operate in a highly controlled, deterministic manner. It first establishes communication with each video camera 50 via the first input / output port 110, initiating a secure session and managing camera configuration and streaming behavior. All packets arriving on the second input / output port from the computer network, and all packets arriving on the first input / output port from the peripheral accessory, are evaluated by the controller against the internal allow-list database before any forwarding occurs. A packet is forwarded across the isolation boundary if and only if its combination of source address, destination address, port, and protocol matches a permitted entry in the allow-list database. The peripheral accessory may initiate communication autonomously — for example, a camera may begin streaming video, push event notifications, or attempt to reach a cloud service — and the controller's forwarding decision is made solely by reference to the allow-list database, independently of whether any prior communication occurred from the computer network side, such as a video server 10, a user observation terminal 20, or a NAS storage 30.

[0050] In some embodiments, the controller may enforce strict traffic filtering policies. The controller evaluates all packets originating from the peripheral accessory against the allow-list database. Packets whose destination address, port, and protocol match a permitted entry are forwarded to the computer network; all other packets are blocked and logged. This applies equally to autonomously initiated communications from the peripheral accessory — such as camera-initiated video streams, event push notifications, cloud service registrations, or firmware self-update attempts — and to communications initiated by the computer network side. The allow-list database is the sole arbiter of what is permitted; the direction of initiation and the presence or absence of a prior command from the network are notfactors in the forwarding decision. Attempted connections to destinations not present in the allow-list database, such as unauthorized outbound tunnels, broadcast packets beyond permitted ranges, or connections to remote systems not approved by the operator, are immediately blocked and logged. Moreover, the controller may ensure that any valid video stream is only routed to the designated computing resource authorized to receive it. This prevents lateral propagation of device-originated traffic within the network and minimizes the risk of exfiltration or unauthorized redistribution.

[0051] The internal allow-list database of the network isolator 100 is the authoritative policy engine governing all traffic traversal across the isolation boundary. Each entry in the allow-list database defines a set of traffic attributes that identify a class of permitted communication. These attributes may include, without limitation: source IP address or address range expressed as a CIDR subnet; destination IP address or address range expressed as a CIDR subnet; one or more transport-layer port numbers or port ranges for TCP and UDP traffic; network protocol identifier; and time-of-day or day-of-week windows during which the entry is active. A packet is forwarded across the isolation boundary if and only if all of its relevant attributes match the conditions specified in at least one allow-list entry. A packet that matches no entry is blocked unconditionally. Crucially, the allow-list database governs traffic in both directions with equal authority: packets originating from the peripheral accessory and directed toward the computer network are subject to the same database-driven evaluation as packets originating from the computer network and directed toward the peripheral accessory. The peripheral accessory is not required to receive a prior command before it may initiate communication; the isolator's decision to forward or block that communication depends entirely on whether the packet's attributes match a database entry, not on the history of prior exchanges between the two sides. This architecture permits operators to define highly granular, context-aware security policies — for example, permitting a camera to push video only to a specific server on a specific port during business hours, while blocking all other outbound traffic from the same camera at all times — without any requirement that those permitted communications be solicited by the network side.

[0052] The network isolator 100 may include a targeted signal routing mechanism that ensures each video stream originating from a specific peripheral video camera 50 is delivered only to one or more designated computers on the network. This step introduces an important layer of signal segregation and destination control, wherein individual video signals from the plurality of video cameras 50 are not broadcast or sent to arbitrary endpoints on the network, but rather are selectively routed to one or more pre-authorized computing devices explicitly configured to receive them.

[0053] This functionality is particularly critical in sensitive or high-security environments, where information compartmentalization is essential to maintain operational integrity and minimize exposure in the event of a breach. For example, in a military surveillance context, video feeds fromcameras 50 monitoring different zones may be intended for different operational units, and it would be both inefficient and insecure for all streams to be available to all endpoints. By assigning unique routing policies for each video stream, the system can isolate access to camera data, enforcing rolebased or location-based data flows.

[0054] The network isolator's controller plays a central role in this process by maintaining an internal mapping of camera-to-recipient associations that is derived from and consistent with the allow-list database. When any video signal originates from a camera on the first input / output port 110 — whether initiated autonomously by the camera or in response to a connection established by a network node — the controller evaluates the signal against both the allow-list database and the camera-to-recipient mapping before forwarding. The signal is forwarded only if its destination address matches both a permitted allow-list entry and the pre-authorized recipient computer associated with that camera in the mapping. If the destination does not match the mapping — for example, if a compromised camera attempts to send a stream to an endpoint other than its designated recipient — the signal is automatically dropped, blocked, and logged for audit purposes, regardless of whether the destination might otherwise appear in the allow-list database for other purposes.

[0055] This approach also introduces robust defenses against man-in-the-middle attacks and internal data leakage, since the isolator ensures that signals never traverse the network freely or indiscriminately. Even in the case of a compromised camera or an insider threat attempting to reroute traffic, the isolator enforces a strict one-to-one or one-to-few communication pattern, anchored by predefined policies that cannot be overridden by the peripheral devices themselves.

[0056] Furthermore, this targeted signal routing method provides scalability and manageability in large camera deployments. It simplifies network architecture by enabling centralized management of camera-to-recipient relationships, reduces unnecessary network traffic, and ensures compliance with organizational data policies or legal frameworks governing video surveillance data distribution.

[0057] Another important aspect of operating the network isolator 100 for improving network security is the capability to inspect the signal generated by a peripheral accessory, such as a video camera or sensor, and remove any unauthorized data, commands, or anomalous content before that signal is allowed to enter the trusted portion of the computer network. This inspection process serves as yet another security enforcement point, effectively transforming the network isolator 100 into a content-aware security filter that sits between potentially untrusted devices and high-value network resources.

[0058] The inspection process begins with the network isolator 100 receiving the raw signal output from the peripheral device via the first input / output port 110. This signal may include the expected primary data payload (e.g., a video stream, image data, or sensor readings), but in many cases, it may also contain metadata, embedded commands, control signals, or other data types, some of which maybe benign, and others potentially malicious or unintended. These extraneous components might be the result of misconfigured firmware, hidden backdoors, unauthorized telemetry, or even deliberate obfuscation introduced during manufacturing.

[0059] Once the signal is received, the network isolator’s internal controller may perform a comprehensive signal parsing and deep packet inspection (DPI). This includes decoding the signal format, identifying standard versus non-standard protocol layers, and isolating the core content from any surrounding or encapsulated commands. During this stage, the network isolator 100 may compare the contents of the signal against a known whitelist of acceptable data formats, structures, and behaviors, as well as a blacklist of suspicious patterns, such as command-and-control messages, unexpected outbound requests, or device self-update instructions. Machine learning-based anomaly detection or rule-based validation may also be applied to identify signals that deviate from normal operational baselines.

[0060] Any portion of the signal that fails validation, such as unauthorized configuration commands embedded within the data stream, covert signaling packets, or metadata indicating attempts to reach out to external destinations, may be stripped from the signal before it is relayed through the second input / output port 150. In cases of significant deviation or clearly malicious content, the isolator may choose to quarantine the entire signal, halt communication with the peripheral, and trigger alerts to system administrators for investigation.

[0061] This process ensures that only clean, validated content whose outer packet attributes have already been cleared by the allow-list database is permitted to traverse the isolation boundary. The peripheral accessory may generate traffic autonomously or in response to network-side connections; in either case, the content sanitization stage applies identically, stripping unauthorized embedded content from any packet that passes the allow-list check before that packet is forwarded across the isolation boundary. The removal of unauthorized elements neutralizes potential attack vectors embedded in peripheral-originated signals — including hardware-trojan-inserted covert channels, firmware-level backdoor payloads, and unauthorized telemetry — and prevents the propagation of these threats to network resources that would otherwise be unreachable by a conventional firewall or endpoint protection tool operating only on packet headers.

[0062] Additionally, this inspection and sanitization process enhances the integrity and reliability of downstream systems, such as video analytics platforms or control centers, by ensuring they are fed with predictable, validated, and safe inputs. It allows organizations to make operational decisions based on trustworthy data while maintaining a secure and compartmentalized network environment.

[0063] The present invention supports a comprehensive approach to security and can be integrated into a “zero-trust security model”, where no device or user, regardless of location or manufacturer, is assumed to be secure by default. In a zero-trust framework, each device must authenticate its identityand demonstrate behavioral compliance continuously, not just at the time of onboarding.

[0064] In further embodiments, identity management solutions, such as public key infrastructures (PKIs), digital certificates, and cryptographic signatures, can be used in tandem with behavioral controls and network isolators described above to continuously validate device trustworthiness throughout its operational life.EXAMPLES OF SPECIFIC TECHNICAL IMPLEMENTATIONS AND ALTERNATIVE EMBODIMENTS

[0065] This section supplements the foregoing description by providing detailed technical implementation specifications and alternative technical solutions for the network isolator described herein. The specifications in this section are based on a preferred commercial embodiment of the invention intended for sale to governments and large enterprises worldwide, and are designed to comply with international security certification standards including FIPS 140-3 Level 3, PSA Certified Level 3, and SESIP (Security Evaluation Standard for loT Platforms) Level 3. Where conflicts between a high-level product summary and a detailed implementation description are identified, those conflicts are noted and the resolution is stated. Novel technical innovations that distinguish the present invention from conventional network address translators and firewalls are identified and described in detail.HARDWARE PLATFORM ARCHITECTURE

[0066] In a preferred embodiment, the network isolator 100 is implemented on a dedicated embedded hardware platform centered on a high-performance, security-oriented microcontroller. In one specific embodiment, the STM32H5 Series microcontroller (MCU) from STMicroelectronics is employed as the central processing unit. This processor family is chosen for its combination of ARM Cortex-M33 compute performance, integrated dual-bank flash, hardware-enforced Arm TrustZone, a hardware memory protection unit (MPU), and an on-chip True Random Number Generator (TRNG) with NIST SP 800-90B certification.

[0067] The dual Ethernet connectivity required for the network isolator is realized as follows: the WanPort (second input / output port 150), which connects to the local area network and potentially to a wide-area network, utilizes the CPU-internal Ethernet MAC; the CamPort (first input / output port 110), which connects to the isolated peripheral accessory subnet, utilizes an external Ethernet controller such as the Microchip LAN9250. This architecture ensures that the two Ethernet interfaces are serviced by separate silicon paths with no shared frame-buffer memory, providing the hardwarelevel separation required by the invention.

[0068] The STM32H5 platform incorporates the following security-relevant hardware features, all of which may be leveraged by the network isolator firmware: (a) an on-chip cryptographic accelerator (CRYP) supporting AES-256-GCM, TDES, and related modes; (b) a hardware hash accelerator(HASH) supporting SHA-256 and SHA-512; (c) a Public Key Accelerator (PKA) for elliptic-curve operations including ECDSA and ECDHE; (d) OTP fuses for immutable device identity anchoring; (e) secure boot with hardware root-of-trust via the Option Byte Loader; (f) an Independent Watchdog Timer (IWDG) for autonomous hardware reset on firmware hang; and (g) a secure non-volatile storage interface for encrypted flash operations. All cryptographic operations within the network isolator SHALL use the hardware accelerators rather than software implementations where available, to minimize attack surface and maximize performance.

[0069] The network isolator firmware executes on the FreeRTOS real-time operating system, using the IwIP TCP / IP network stack for protocol processing. FreeRTOS is selected for its deterministic scheduling, well-characterized memory footprint, and native MPU integration, enabling task-level memory isolation. Separate FreeRTOS tasks are allocated for: (1) the NAT and packet forwarding engine; (2) the administrative web server; (3) the camera-discovery state machine; (4) the whitelist enforcement module; (5) the logging subsystem; and (6) system health and watchdog monitoring. These tasks communicate exclusively through FreeRTOS message queues and semaphores, with MPU configurations preventing any task from directly accessing the memory regions of other tasks.

[0070] Alternative hardware embodiments of the present invention may employ other embedded processor platforms with comparable security capabilities, including but not limited to: NXP i.MX RT series microcontrollers with hardware cryptography and Ethernet support; Renesas RZ / N2 series industrial networking processors; Marvell ARMADA embedded networking SoCs; or any ARM Cortex-M or Cortex-A class processor with a hardware random number generator, hardware cryptography accelerators, and hardware memory protection. In such alternative implementations, the dual-Ethemet architecture, NAT engine, whitelist enforcement, physical isolation guarantees, and all security policies described in this specification remain fully applicable, subject to adaptation of the firmware to the target hardware platform.

[0071] For production deployments, the following security hardening measures SHALL be applied at the hardware level: JTAG / SWD debug interfaces shall be permanently disabled via OTP option bytes; the boot ROM shall be configured to verify firmware signatures before execution; all unused GPIO pins shall be configured as inputs with pull-downs to prevent floating-pin attacks; and unused peripheral clocks shall be gated off at boot time to minimize the device's electromagnetic and power side-channel signature.One-Camera-Per-Unit Architecture and Network Address Translation Engine

[0072] In the preferred commercial embodiment, each network isolator 100 is configured to support exactly one IP-enabled peripheral accessory — specifically one IP camera — per physical unit. Organizations requiring protection for multiple cameras deploy one network isolator per camera in parallel. This one-camera-per-unit constraint: (a) simplifies the security proof by reducing theisolation model to a single untrusted device per unit; (b) ensures predictable resource allocation within the NAT session table and the whitelist enforcement engine; (c) facilitates independent security certification per unit; and (d) eliminates the risk that a compromised camera on one port can communicate with another camera on another port of the same device.

[0073] Conflict Notice: The Summary of the Invention and the claims as filed in the underlying U.S. application (Docket BK-13) contemplate configurations in which a plurality of peripheral accessories, including a plurality of video cameras, may be operatively connected to a single network isolator. The Summary-level multi-camera capability is the broader, claiming scope of the invention and takes precedence as the inventive concept. The one-camera-per-unit constraint is a preferred implementation decision for the commercial product. The invention is not limited to single-camera operation; multi-camera embodiments within a single isolator unit remain within the scope of the claims. The detailed implementation described in this section represents a specific, preferred, commercially optimized embodiment.

[0074] The network isolator performs symmetric Network Address Translation (NAT) to relay authorized traffic between the CamPort and WanPort transparently, so that neither the camera nor nodes on the WAN are aware of the isolator's presence in the traffic path. The NAT implementation maintains a session table of up to 256 concurrent sessions. Of these, 16 entries are reserved for device-originated control-plane traffic (DNS resolution, NTP synchronization, syslog forwarding, firmware update checks, SMTP notifications, LDAP / RADIUS authentication, and administrative services), leaving 240 session entries available for camera traffic. The maximum size of the session table is configurable in the firmware source code. When the session table for camera traffic is full, the oldest idle session is evicted; if no sessions are idle, the new connection is rejected and the event is logged.

[0075] NAT session entries are created exclusively by outbound traffic originating from the camera. Unsolicited inbound traffic that does not match an existing NAT session entry is dropped immediately. Per-protocol session idle timeouts are: TCP established sessions — 86,400 seconds; TCP transitory (SYN / FIN / RST) — 60 seconds; UDP sessions — 180 seconds; ICMP — 30 seconds. All timeout values are configurable in firmware source code. An alternative embodiment may employ connection tracking with stateful inspection to further reduce the session table size by consolidating bidirectional flows.

[0076] The WanPort defaults to DHCP client mode. If no DHCP response is received within ten seconds, the isolator falls back to automatic IP address assignment in the link-local range (169.254-x.x) per RFC 3927. The isolator advertises its administrative interface via rnDNS (RFC 6762) as "videoisolator-<last-four-digits-of-WAN-MAC>.local" exclusively on the WAN port to enable zero-configuration discovery by installers; this rnDNS advertisement is a control-plane-only service generated by the administrative subsystem and is never forwarded to or detectable from theperipheral-accessory side of the isolation boundary. The full dual-plane architecture governing this advertisement is described in the Dual-Plane Transparent Architecture section below. If the WAN IP address changes during a DHCP lease renewal, all active NAT sessions are reset, the event is prominently logged, and the self-signed TLS certificate for the administrative interface is regenerated to include the new WAN IP as a Subject Alternative Name.

[0077] An alternative embodiment may employ a transparent bridging mode rather than NAT, in which the isolator operates at Layer 2, forwards Ethernet frames based on a whitelist of permitted MAC-to-MAC flows, and performs deep packet inspection without modifying IP addresses. This approach eliminates the NAT compatibility issue with camera cloud services that require stable external port mappings, at the cost of exposing the camera's MAC address and IP address to the WAN. A further alternative embodiment may employ a VLAN-based segmentation architecture, in which the camera is placed on a dedicated VLAN and the isolator acts as the single authorized interVLAN router with whitelist enforcement.Camera Discovery, Address Configuration, and Network Binding

[0078] A key operational challenge for non-technical installers is correctly configuring the IP addressing relationship between the network isolator and the attached camera. The present invention addresses this challenge through an automatic, multi-method camera discovery system that concurrently runs three discovery pathways against the isolated CamPort subnet and atomically commits on whichever method first identifies the camera, cancelling all other methods in the same atomic operation. This atomic first-wins multi-method discovery eliminates race conditions that could arise if two discovery methods simultaneously identified conflicting addresses.

[0079] The three concurrent discovery methods are: (1) Embedded DHCP Server — the network isolator activates an internal DHCP server on CamPort and awaits a DHCP request from the camera. If a DHCP request is received, the isolator assigns the camera a default IP address of 192.168.0.100 with a default gateway of 192.168.0.101 (the CamPort address). Both the camera IP address and the CamPort address are configurable by the user through the administrative interface before discovery occurs; after discovery, these settings become read-only until the camera is disconnected or a manual reset is performed. (2) Passive ARP Observation — if no DHCP request is received within a configurable timeout period (default 30 seconds from CamPort link-up detection), the isolator passively monitors the CamPort for ARP packets and uses the Sender Protocol Address (SPA) from the first ARP request header as the camera's static IP address. (3) Manual Entry — the administrator may manually enter the camera's IP address through the administrative web interface at any time. Manual entry atomically cancels the two automatic methods.

[0080] After the discovery timeout, if no camera IP address has been determined by any method, the isolator continues listening for DHCP or ARP packets indefinitely, retrying the detection scan every30 seconds. The administrative interface displays a "waiting for camera" status indicator. If an administrator is present at the interface when the timeout expires, the interface prompts for manual IP entry with a clear on-screen explanation. A note in the user interface informs the installer that some cameras may require a timeout period longer than the default 30 seconds before initiating network communication.

[0081] When the camera's IP address is determined by ARP-based or manual discovery (i.e., the camera has a pre-configured static IP address), the network isolator must assign itself a CamPort IP address on the same subnet as the camera. The isolator selects its CamPort IP address using an adaptive algorithm with ARP-safety probing: (a) the isolator first probes the camera's subnet ".101" host address (for example, if the camera is at 10.0.1.50, the isolator probes 10.0.1.101) by sending an ARP probe and waiting for a response; (b) if the ".101 " address is in use (including being used by the camera itself), the isolator falls back to the ".254" address on the same subnet; (c) before assigning any address, the isolator verifies via ARP probe that the address is unused; and (d) if the ARP probe on the selected address receives a response indicating another device is present on the CamPort subnet besides the camera, the isolator refuses to proceed and displays an error to the administrator, because only one device (the camera) is permitted on the CamPort subnet.

[0082] After camera discovery is complete, the network isolator performs a comprehensive ARP scan of the CamPort subnet to verify that exactly one IP address is present on that subnet. If more than one IP address responds to the ARP scan, the isolator treats this as a security error, displays an alert to the administrator, and refuses to forward camera traffic until the extraneous device is removed. This ARP scan is repeated at startup and after any configuration change. If a new device is detected on CamPort after initial setup, an alert is displayed in the administrative interface and optionally forwarded to the configured syslog server.

[0083] Once camera discovery is complete, the network isolator binds the discovered camera MAC address to its IP address and rejects any subsequent ARP packet on CamPort that would change this binding. This ARP-binding lock prevents ARP spoofing attacks on the isolated camera subnet. Additionally, if the CamPort Ethernet link goes down and then comes back up, the isolator automatically re-runs the full camera discovery process, including ARP scanning and MAC binding, to detect whether a different physical device has been connected to the CamPort cable. The link-down / link-up event and any resulting discovery results are logged. This mechanism defends against "silent camera swap" attacks in which an attacker physically replaces the camera with a different device between the camera's cable and the CamPort.

[0084] An additional layer of link-layer protection is provided by CamPort source-MAC pinning: the network isolator drops any Ethernet frame arriving on CamPort whose source MAC address does not match the bound camera MAC address, regardless of IP-layer address matching. This belt-and-suspenders defense operates at Layer 2 and is independent of the ARP binding mechanism, providing protection even against attacks that operate below the IP layer.

[0085] Alternative embodiments of the camera discovery mechanism include: (a) ONVIF device discovery using the WS-Discovery protocol, in which the isolator sends a WS-Discovery Probe message on the CamPort and identifies the camera from the resulting ProbeMatch response; (b) mDNS / DNS-SD discovery, in which the isolator listens for mDNS service announcements from the camera; (c) RTSP DESCRIBE probing, in which the isolator sends an RTSP DESCRIBE request to known RTSP port ranges and identifies the camera from the response; and (d) manufacturer-specific discovery protocols for specific camera vendors. These alternative discovery methods may be used as fallback mechanisms after the primary three-method discovery system.Whitelist-Based Access Control and Traffic Filtering

[0086] The core security mechanism of the network isolator is a whitelist-based access control system that governs all traffic to and from the connected peripheral accessory. No traffic to or from the camera is permitted unless it matches an entry in the whitelist. Each whitelist entry specifies: (a) an IP address or CIDR subnet; (b) one or more TCP / UDP port numbers, a range of ports, or an "all ports" flag; and (c) optionally, a MAC address for nodes on the same subnet as the WanPort. For nodes on a different subnet than the WanPort (remote nodes), MAC address matching is not applicable and the MAC field is not displayed or is marked as "N / A (remote)" in the administrative interface. The whitelist supports a maximum of 128 user-defined entries, and remaining capacity is displayed in the administrative interface. System-managed entries (DNS servers, NTP servers, syslog server, SMTP server, firmware update server, LDAP / RADIUS authentication servers) are maintained in a separate internal whitelist not counted against the user entry limit.

[0087] The administrative interface presents the whitelist in plain-language form accessible to nontechnical users. Rather than displaying raw IP:Port:MAC tuples, the interface presents two categorized views: "Who can view this camera?" (a list of allowed viewer nodes by name or IP) and "What cloud services can the camera use?" (a list of allowed cloud service destinations, classified by vendor and service type where recognizable). An "Advanced" toggle reveals the underlying technical tuples for expert users. Blocked nodes are displayed in red and allowed nodes in green to provide immediate visual distinction.

[0088] All traffic between the camera and any node not present in the whitelist is blocked and logged. For each blocked connection attempt, the log records the remote IP address, the port number, the MAC address (for same-subnet nodes), the direction (camera-to-WAN or WAN-to-camera), a count of repeated attempts from the same source, and if the connection attempt involved a public IP address, the result of a reverse DNS lookup of that address. An administrator reviewing the log may click on any blocked node entry to open a pre-filled "Add Allowed Device" dialog containing the blockednode's IP, port, and MAC address (if applicable), enabling one-click addition of the blocked node to the whitelist.

[0089] Per- whitelist-entry rate limiting is enforced independently for each entry, with a default limit of 10 Mbps per entry regardless of whether the entry uses explicit ports or the "all ports" flag. Rate limits are configurable per entry through the administrative interface. The "all ports" convenience does not bypass rate limiting. When a flow is dropped due to rate limiting, the syslog entry distinguishes between "per-entry cap hit" and "global CamPort cap hit" to enable operators to tune the correct parameter. An administrator may increase the per-entry rate limit for high-bandwidth applications such as 4K video streams.

[0090] ICMP handling is separate from the port-based whitelist rules. ICMP echo (ping) between the isolator and the camera on CamPort is always permitted for health-check purposes. ICMP echorequest and echo-reply from WanPort to the isolator are permitted only from whitelisted nodes. ICMP redirect, source quench, and all other ICMP types are dropped unconditionally. The isolator also drops all IPv6 traffic on both ports, aggregating IPv6 attempt counts in the log to prevent log flooding. Multicast and broadcast traffic from the camera to WanPort is blocked by default; only unicast traffic matching whitelist entries may traverse the isolation boundary.

[0091] Alternative access control embodiments include: (a) role-based access control (RBAC), in which different users or groups of users may be granted access to different cameras or camera streams; (b) time-based access control, in which whitelist entries are valid only during specified time windows; (c) geolocation-based access control, in which the source IP address is mapped to a geographic region and access is restricted to specified regions; and (d) cryptographic identity-based access control, in which cameras and viewers authenticate using X.509 certificates and access is granted based on certificate attributes rather than IP addresses.Dual-Plane Transparent Architecture with Out-of-Band Administrative Access

[0092] A fundamental architectural feature of the network isolator 100 is the simultaneous operation of two completely independent traffic planes that share a single network-side IP address, enabling the isolator to remain fully transparent in the forwarding path while maintaining an independently accessible administrative interface that is invisible to the peripheral accessory and to forwardingplane traffic on the computer network.

[0093] The forwarding plane comprises the NAT engine and the allow-list enforcement module. All packets arriving on either the first input / output port 110 or the second input / output port 150 that match a permitted allow-list entry are processed by the forwarding plane. From the perspective of a node on the computer network, the peripheral accessory appears to be reachable directly at the network isolator's WAN IP address; the presence of the isolator in the traffic path is not detectable because the isolator does not insert itself as a distinct IP hop — it modifies source and destinationaddresses via NAT so that the camera and the network node exchange packets as if communicating directly. From the peripheral accessory's perspective, the isolator's CamPort IP address is merely a default gateway; the accessory has no knowledge of the network isolator as anything other than a router providing connectivity to the network.

[0094] The invisibility of the network isolator to both sides of the isolation boundary is itself a security property, distinct from the traffic filtering and content sanitization functions described elsewhere in this specification. A peripheral accessory that can detect the presence of a NAT gateway or security appliance between itself and the network gains topological intelligence about the security infrastructure protecting it. A compromised peripheral equipped with hardware-embedded malicious components may use this topological information to adapt its attack strategy — for example, by timing communications to coincide with expected polling intervals, by probing for management interfaces on the gateway's own address, or by crafting traffic specifically designed to evade the gateway's detection logic. The network isolator defeats this class of reconnaissance by ensuring that neither the peripheral accessory nor the network-side nodes can distinguish the isolator from an ordinary default gateway or router. From the peripheral accessory's perspective, the CamPort IP address is simply a default gateway with no distinguishing characteristics. From the network-side perspective, the camera appears to be a device directly reachable at the WAN IP address with no intermediate security appliance. This mutual invisibility means that a compromised peripheral cannot determine what security controls are in place, cannot identify the make, model, or capabilities of the security device, and cannot target the security device's administrative interface, management ports, or configuration system from the camera side.

[0095] The control plane comprises the administrative web server, the device configuration database, the allow-list management system, the authentication subsystem, and all device-management services. The control plane is accessible exclusively via a reserved administrative port — by default port 3854 — on the WAN IP address of the network isolator. Crucially, this reserved port is explicitly excluded from the NAT forwarding rules. A packet destined for the WAN IP address on the reserved administrative port is never forwarded to the peripheral accessory; it is instead terminated at the network isolator itself and served by the administrative web server. The peripheral accessory has no IP address, no Ethernet path, and no protocol mechanism by which it can reach the administrative control plane, because the administrative control plane does not exist on any address or port accessible from the first input / output port 110.

[0096] This architecture gives the network isolator a dual identity at the single WAN IP address: on permitted forwarding ports, the WAN IP represents the peripheral accessory as proxied by the NAT engine; on the reserved administrative port, the same WAN IP represents the network isolator's own administrative interface. These two identities are completely isolated from each other at every layerof the protocol stack, enforced by the same task-level and memory-level separation described in the Hardware Platform Architecture section above.

[0097] The port-discriminating traffic classifier is the component responsible for assigning each incoming WAN-side connection to either the forwarding plane or the control plane. Its decision algorithm operates as follows. For every new TCP connection or UDP packet received on the second input / output port 150, the classifier first examines the destination port number. If the destination port matches the configured administrative port (default 3854), the connection is unconditionally routed to the control plane and served by the administrative web server; it is never evaluated by the NAT forwarding engine and is never forwarded to the peripheral accessory, regardless of whether the source node is present in the allow-list database. If the destination port does not match the administrative port, the classifier examines the source IP address and port against the allow-list database. If the source is a whitelisted node and the destination port is among the ports permitted for that entry, the connection is routed to the forwarding plane and the NAT engine transparently proxies it to the peripheral accessory; the peripheral accessory receives the connection as if originating directly from the network-side node. If the source is not whitelisted or the destination port is not permitted for the source, the connection is blocked and logged; for HTTPS connections on port 443 from non-whitelisted sources, the isolator completes the TLS handshake using its self-signed certificate and returns an HTTP 302 redirect to the administrative interface URL (https: / / <WAN-IP>:3854 / ), allowing the connecting party to reach the administrative interface without knowing the administrative port in advance. For HTTP connections on port 80 from any source, the isolator returns an HTTP 301 redirect to the HTTPS administrative interface URL. This redirect behavior means that any browser-based access attempt to the isolator's WAN IP address by a non-whitelisted node results in transparent guidance to the administrative interface rather than a connection error, improving installer usability without compromising security. In all cases, the administrative port 3854 is served to any connecting party regardless of allow-list status, because restricting administrative access by IP address would prevent initial configuration before any allow-list entries exist; access control for the administrative interface is enforced instead by the password and optional two-factor authentication described in the Authentication section. The administrative port number and the HTTPS redirect behavior are both configurable. It is noted that this port-based discrimination is one embodiment of a more general WAN-port multiplexing architecture. An alternative embodiment employs IP-address-level multiplexing: the second input / output port is configured with two IP addresses, a first serving exclusively the management function and a second serving exclusively the pass-through forwarding function. In this alternative, the traffic classifier first examines the destination IP address; if it matches the management IP address, the packet is unconditionally assigned to the control plane regardless of destination port; if it matches the forwarding IP address, the packet is evaluated againstthe allow-list database for pass-through forwarding. A combined embodiment using both IP-address and port-based discrimination is also contemplated, providing the most granular multiplexing control.

[0098] The security advantage of this out-of-band administrative architecture is significant. A peripheral accessory that has been compromised — including one containing hardware-embedded malicious components — cannot discover, probe, enumerate, or attack the administrative system of the network isolator, because that system is not reachable from the peripheral-accessory side of the isolation boundary by any means. Even if a compromised peripheral accessory were to send packets to the CamPort IP address of the isolator, those packets would be evaluated against the allow-list database and dropped if they do not match a permitted entry — and the allow-list contains no entry permitting peripheral-originated traffic directed to the isolator's own administrative system. This is a qualitatively stronger protection than that offered by a conventional out-of-band management interface, because in the present invention the management interface is not on a separate physical interface requiring additional cabling — it is logically out-of-band while remaining physically in-band on the same WAN port, making the architecture deployable without any additional network infrastructure.

[0099] For initial setup and zero-configuration discovery, the network isolator advertises its administrative interface via mDNS (Multicast DNS, RFC 6762) on the WAN port, using the hostname "videoisolator-<last-four-hex-digits-of-WAN-MAC>.local" as a Subject Alternative Name in its self-signed TES certificate. This mDNS advertisement is generated by the control plane and is entirely absent from the forwarding plane — the peripheral accessory on CamPort never sees or hears the mDNS traffic. An administrator connecting a laptop or mobile device to the same LAN segment as the WAN port can therefore discover and configure the network isolator without knowing its IP address, without any pre-configuration, and without affecting or interrupting the forwarding plane's operation.

[0100] In an alternative embodiment, the reserved administrative port number is user-configurable, enabling deployment in environments where port 3854 conflicts with other services. In a further alternative embodiment, the administrative interface may be completely disabled via the administrative UI once the device is configured, after which it is inaccessible from the network and can only be re-enabled by a physical button press on the device. In this locked-down configuration, the WAN IP address serves exclusively as the camera's NAT address, and the isolator becomes completely invisible to all network-side traffic. In yet a further alternative embodiment, the network isolator may support a dedicated out-of-band management physical port separate from the WAN port, in which case the administrative interface is accessible only via that dedicated management port and is entirely absent from the WAN IP address, providing the strongest possible separation between the forwarding and control planes.Administrative Interface and User Experience

[0101] The network isolator provides a web-based administrative interface accessible via HTTPS on port 3854 (configurable). The administrative interface is accessible exclusively from the WanPort; the CamPort serves no administrative functions. The interface is designed to be usable by nontechnical users, including field installers without network security expertise. All user- visible text avoids technical jargon: "Allowed Devices" is used instead of "whitelist"; "Camera Network Port" instead of "CamPort"; "Network Port" instead of "WanPort." Internal firmware code and API parameters use the technical names.

[0102] The administrative interface supports the following display languages: English (default), Spanish, French, German, Chinese (Simplified), Japanese, Hebrew, Arabic, Hindi, and Indonesian. Language selection is stored in the device configuration. On hardware-constrained embedded systems, localization is implemented using a compressed key-value string table with deflate compression, decompressing only the selected language at runtime to minimize memory usage.

[0103] The administrative interface is implemented as a responsive web application compatible with mobile devices, tablets, and desktop browsers, enabling field installers to configure the device using a smartphone. The interface uses Server-Sent Events (SSE) or WebSockets to push real-time status updates — including blocked access attempts, camera disconnection events, and NAT session counts — to the browser without requiring manual page refresh. The interface serves strict Content Security Policy headers (CSP) preventing XSS attacks, including "default-src 'self; script-src 'self; style-src ’self", and HTTP Strict Transport Security (HSTS) headers when a CA-issued certificate is installed.

[0104] The camera status dashboard presents a real-time summary including: camera connection status with a color-coded indicator (green for connected, amber for disconnected); WAN connectivity status; count of allowed devices with remaining capacity; count of blocked access attempts in the last 24 hours; current NAT session count; and a graphical diagram of the form [Camera] — [Isolator] — [Network] with green / red indicators for each link. Active NAT sessions and current whitelist connections are displayed to allow the administrator to verify that expected traffic is flowing.

[0105] On error conditions, the administrative interface provides guided recovery instructions in plain language rather than technical error codes. When the CamPort link is down, the interface displays "No camera detected. Please check the Ethernet cable connection" with a cabling diagram. When camera traffic from an unauthorized node is detected, the interface provides contextual action buttons allowing the administrator to add the node to the allowed devices list or view detailed connection information.

[0106] An alternative administrative access model may employ an out-of-band serial console interface for initial configuration and emergency recovery, supplementing or replacing the web-basedinterface in high-security environments where remote administrative access is prohibited.Setup Wizard and Non-Forwarding Learning Mode

[0107] To guide non-technical users through the initial configuration of the network isolator, the device provides a step-by-step setup wizard comprising the following steps: Step 1 — Change the factory-default password and set a new administrator password; Step 2 — Connect the camera to the CamPort and allow the automatic camera discovery process to identify the camera (or enter the camera IP address manually); Step 3 — Run the Learning Mode (described below); Step 4 — Review and approve the suggested whitelist entries presented by the learning mode; Step 5 — Optionally install a custom TLS certificate for the administrative interface; Step 6 — Test camera access using the built-in "Test Camera Access" button to verify end-to-end connectivity; Step 7 — Complete setup and begin normal operation. The wizard displays a progress indicator (e.g., "Step 3 of 7") so installers can track their progress.

[0108] The Learning Mode is a time-bounded non-forwarding monitoring state in which all camera-originated outbound traffic is captured and recorded as whitelist candidates but not forwarded to the WAN. This mode runs for a configurable period with a default of five minutes and a maximum of ten minutes. During Learning Mode, the camera's TCP SYN packets, UDP first-packets, DNS queries, IGMP messages, and all other outbound communication attempts are intercepted and recorded by the network isolator without being forwarded. This allows the camera to reveal the complete set of external servers and services it needs to access, without any of that communication actually occurring on the WAN.

[0109] Critically, during Learning Mode, the network isolator's DNS proxy remains active and operational. This means that DNS queries from the camera are resolved by the isolator and the hostname-to-IP correspondence is recorded alongside the destination IP address and port. When the Learning Mode period ends, the administrator is presented with a categorized list of the captured connection attempts in plain language, for example: "hikvision-update.example.com (185.x.x.x):443 — HTTPS cloud connection" rather than merely "185.x.x.x:443." This enables meaningful whitelist authoring decisions without requiring the administrator to possess network expertise.

[0110] The Learning Mode state machine implements power-loss-safe operation through a transaction log. If the device loses power during Learning Mode, on the next boot the device recovers the partial capture state from the transaction log and presents the captured destinations to the administrator. The Learning Mode is exited atomically: when the timer expires or the administrator manually ends the mode, all captured destinations are presented simultaneously and the temporary monitoring state is destroyed, including any associated buffer memory. After the administrator reviews and approves the suggested whitelist entries, the camera is expected to retry its connection attempts, which will now succeed because the corresponding whitelist entries have been added.

[0111] The network isolator may attempt to identify the connected camera's manufacturer using a three-method hybrid detection approach during the Learning Mode: (1) MAC address Organizationally Unique Identifier (OUI) lookup at link-up, providing an immediate vendor identification before Learning Mode begins; (2) DNS query pattern matching during Learning Mode, in which queries containing recognizable vendor domain names (e.g., "Lhikvision.com", "Ldahuasecurity.com") are matched against a vendor signature database; and (3) destination IP range matching, in which the camera's attempted outbound connections are matched against known IP ranges associated with specific camera vendors. The combined identification from all three methods is used to pre-fill whitelist entries from a static, firmware-resident preset database covering up to ten major camera brands including Hikvision, Dahua, Axis, Bosch, Hanwha, Vivotek, Uniview, Mobotix, ACTi, and generic ONVIF-compliant cameras. Adding additional camera presets to the database requires a firmware update.Authentication, Cryptography, and Security Architecture

[0112] The network isolator implements a multi-layer authentication and session security architecture. There is a single administrator account. The default password is unique per device, generated from a device-unique identifier or a random value assigned at manufacturing, and is printed on a label affixed to the device. Universal or shared default passwords across multiple devices are explicitly prohibited. Upon first login with the factory-default password, the administrator is required to change the password before any other configuration is accessible. The new password must be a minimum of eight characters and include at least one uppercase letter, one lowercase letter, and one digit. The last twelve password hashes are stored and re-use of any of the twelve most recent passwords is prohibited. The number of stored hashes is configurable in firmware source code.

[0113] Passwords are stored as salted hashes using bcrypt with a cost factor of 12 or higher on the target hardware platform. The salt is at least 16 bytes derived from the hardware TRNG. Plaintext passwords are never stored or logged. To avoid the known bcrypt 72-byte silent truncation issue, passwords longer than 72 bytes are pre-hashed with SHA-256 before the bcrypt operation. Constanttime comparison functions are used for all security-sensitive comparisons including session tokens, HMAC values, password hash verification, and certificate comparison, to prevent timing sidechannel attacks.

[0114] The administrator may optionally configure Two-Factor Authentication (2FA) using Time-Based One-Time Passwords (TOTP) per RFC 6238. During TOTP enrollment, the device generates a QR code and text secret displayed on the administrative web interface. The administrator scans the QR code with an authenticator application (such as Google Authenticator or Microsoft Authenticator) and enters a valid TOTP code to confirm enrollment. Five single-use recovery codes are generated and displayed once during enrollment. The TOTP secret is immediately wiped from RAM and fromthe rendered HTML buffer after the administrator confirms the first valid TOTP code. Recovery codes are similarly displayed once and never again. TOTP secrets stored in flash are encrypted using AES-256-GCM with a key derived from the CPU's unique device identifier.

[0115] The administrator may also configure external authentication using LDAP (with LDAPS or STARTTLS, RFC 4513) or RADIUS (RFC 2865) as an alternative authentication backend for the single administrator account. When external authentication is configured, the administrator's credentials are validated against the configured external server. A local password authentication fallback remains available and cannot be disabled, activated automatically when the external authentication server is unreachable. When LDAP / RADIUS is configured, TOTP (if enabled) serves as the second factor applied after successful primary authentication regardless of the primary method. The external server's TLS certificate must be validated against a CA certificate uploaded by the administrator; self-signed server certificates are supported via explicit CA upload. RADIUS authentication requires RFC 5080 / 5997 Message-Authenticator on every Access-Request and Access- Accept to prevent forgery. Plaintext LDAP (without TLS) is not supported.

[0116] Administrative web sessions use 128-bit session tokens generated by the hardware TRNG. Sessions are bound to the client's IP address at authentication; requests arriving with a valid session cookie from a different IP address cause immediate session invalidation. Session tokens are rotated every five minutes during active sessions. The idle session timeout defaults to 15 minutes and is configurable. Only one concurrent administrative session is permitted; if a new administrator logs in, the existing session is invalidated. After five consecutive failed login attempts, exponential backoff is imposed (30 seconds, 60 seconds, 120 seconds, etc.). Concurrent in-flight password hash verification operations are capped at two to prevent CPU exhaustion via parallel slow-hash attacks. All failed login attempts are logged.

[0117] The administrative HTTPS interface uses TLS 1.2 or higher with strong cipher suites only (AES-256-GCM with ECDHE key exchange). TLS 1.0 and TLS 1.1 are disabled. A self-signed TLS certificate is generated at first boot with a two-year validity period, carrying Subject Alternative Names for the device's mDNS hostname (videoisolator-<MAC4>.local), the configured WAN IP address, and the literal hostname "videoisolator.local. " The certificate is automatically regenerated when the WAN IP address changes, reducing spurious TLS warning prompts to administrators during normal DHCP lease changes. The administrator may install a custom CA-issued certificate. When a custom certificate is installed, HSTS headers may be enabled. The administrative interface warns the administrator when the certificate is approaching expiration. The administrator may optionally restrict access to the administrative interface to specific IP address ranges, a capability commonly required in government network deployments.

[0118] Government and enterprise deployments requiring certificate-based authentication foradministrators may configure client TLS certificate authentication. The administrator uploads a CA certificate that is trusted to issue administrator client certificates, such as certificates from Common Access Card (CAC) or Personal Identity Verification (PIV) smart card infrastructure. Client certificate trust anchor material is stored encrypted alongside other device secrets.DNS Proxy, NTP Proxy, and Protocol Filtering

[0119] To prevent DNS tunneling and unauthorized data exfiltration via DNS queries from the camera, the network isolator implements a transparent DNS proxy that intercepts all DNS queries (port 53, UDP and TCP) originating from the camera on CamPort, regardless of the destination IP address specified in the query. The proxy forwards the query to configured upstream DNS resolvers (defaulting to Google Public DNS at 8.8.8.8 and 8.8.4.4) and returns the response to the camera. The camera is never permitted to communicate directly with any DNS server; all DNS traffic is mediated by the isolator's proxy. The DNS proxy validates response format, enforces a maximum of ten answer records per response, and strips all record types except A, AAAA, and CNAME from responses before forwarding them to the camera. TXT, NULL, SRV, MX, and other record types that are commonly used for DNS tunneling are blocked at the protocol level.

[0120] Similarly, a transparent NTP proxy intercepts all NTP traffic (port 123, UDP) from the camera and forwards it to configured upstream NTP servers (defaulting to Google Public NTP at 216.239.35.0 and 216.239.35.4). The NTP proxy validates response packet structure and rejects packets that do not match the expected NTP version and mode. This architecture ensures that the camera cannot use NTP as a side-channel while still maintaining accurate time synchronization. An enhanced embodiment may employ Network Time Security (NTS, RFC 8915) for the device's own NTP synchronization, providing authenticated time assurance for all device security functions (certificate validity checks, log timestamps, TOTP window calculations, password expiry) while presenting standard NTP to the camera.

[0121] The upstream DNS and NTP server addresses are maintained in the system whitelist and are configurable by the administrator. The DNS proxy additionally randomizes source port selection for upstream queries to mitigate cache poisoning attacks (per RFC 5452), validates that response transaction IDs match the corresponding request IDs, and enforces response size limits. DNSSEC validation is deferred to a future implementation version.Firmware Security, Secure Boot, and Update Mechanism

[0122] The network isolator implements secure boot using the hardware root-of-trust capabilities of the STM32H5 platform. At every boot, the boot ROM verifies the cryptographic signature of the firmware image before execution begins. The firmware signature uses Ed25519 or ECDSA P-256 digital signatures. If the primary firmware partition fails signature verification, the bootloader automatically attempts to boot from the backup firmware partition. If both partitions fail verification,the device enters a fail-shut state in which neither Ethernet port is brought up, the device is not accessible from the network, and a solid red LED indicates the hardware fault. This fail-shut behavior ensures that a physically compromised device cannot pretend to be functional, a critical property for a device whose core purpose is to enforce a security boundary.

[0123] Firmware is stored in A / B dual-partition format. The active partition and backup partition each contain a complete, independently bootable firmware image. When a firmware update is applied, the new image is written to the inactive partition, its signature is verified, and then the bootloader's active partition pointer is updated. If the new firmware fails to boot correctly (detected by a boot counter that increments before initialization and clears on successful completion of the startup sequence), the bootloader reverts to the backup partition. A monotonic version counter anchored in OTP fuses is included in every firmware image and prevents rollback to any version other than the immediately previous version.

[0124] Firmware updates are sourced from the AKCP update server at a configurable URL. The device authenticates to the update server using a device-unique HMAC-signed request that includes the device serial number, a monotonic counter value, a nonce, and a timestamp. The HMAC key is unique per device and is burned into OTP fuses at manufacturing. The update-check schedule is approximately once per day with a uniform random jitter of ±2 hours derived from the hardware TRNG, preventing synchronized fleet requests that could overload the update server and preventing the creation of a predictable outbound beacon pattern that could be used to identify devices by traffic analysis. All firmware images downloaded from the update server are verified against a cryptographic signature before installation; unsigned or incorrectly signed firmware images are refused. The TLS connection to the update server uses certificate pinning; the pinned CA certificate can only be updated via a signed firmware update.

[0125] The device's own control-plane traffic — including DNS resolution, NTP synchronization, syslog forwarding, SMTP notifications, firmware update checks, and LDAP / RADIUS authentication — originates from the WanPort IP address and bypasses the NAT forwarding engine entirely. An internal egress firewall restricts device-originated WAN traffic to a hardcoded list of permitted protocol and port combinations: syslog (514 / UDP, 6514 / TCP), SMTP (587 / TCP), LDAP (636 / TCP), RADIUS (1812 / UDP), NTP (123 / UDP), DNS (53 / UDP and TCP), and HTTPS (443 / TCP for firmware updates). Furthermore, destination IP addresses for device-originated traffic are restricted to the specific configured server IP addresses, not arbitrary destinations on those ports. This prevents a compromised firmware from exfiltrating data to an attacker-controlled server even via the device's own legitimate services.Logging, Forensics, and Audit Trail

[0126] The network isolator maintains four independent log types, each with its own storage andretention policy: (1) CamPort blocked traffic log — records all traffic originating from the camera that is blocked by the whitelist; (2) WanPort blocked traffic log — records all inbound WAN traffic that is blocked by the whitelist; (3) administrative audit log — records all administrator configuration changes including whitelist additions / removals, password changes, and settings modifications, with timestamp, session ID, and source IP; and (4) security and forensic log — records security-critical events such as tamper detection, authentication failures, firmware updates, and secure boot results.

[0127] Each log type maintains a circular FIFO buffer of up to 1,000 entries stored in flash and RAM. The security and forensic log may optionally be configured as a write-once append-only log; in write-once mode, when the log reaches capacity it stops updating until the administrator explicitly permits it to operate in FIFO mode. The security and forensic log triggers an alarm when it reaches 90% capacity. Repeated connection attempts from the same source to the same destination are aggregated into a single log entry with an incrementing count, rather than creating separate log entries for each attempt, preventing log flooding.

[0128] Fog entries are protected against tampering using a chained HMAC mechanism. Each log entry includes an HMAC of the previous entry, forming a tamper-evident chain. A double-buffer write strategy is used: new entries are written to a standby buffer with their HMAC, and then made active via an atomic pointer swap, ensuring power-loss-safe log appends. On boot, the log chain is verified from the beginning and truncated at the first invalid HMAC; the truncation event itself is recorded as the next entry. Periodic log checkpoints are written to a separate flash sector to prevent a single flash sector failure from invalidating the entire log chain. Administrators may export the audit log as a digitally signed file, with the export including the device serial number, time range, and NTP synchronization status at the time of export, for external verification.

[0129] A write-once flash sector, separate from the FIFO log and not erasable by factory reset or PURGE operation, records critical security events: tamper detection, consecutive authentication failures, firmware update events, and secure boot results. This persistent forensic record survives device compromise and is available for post-incident forensic analysis. Each boot's secure boot result, including the firmware version and hash, is written to this write-once sector.

[0130] The network isolator supports syslog forwarding per RFC 5424 to an externally configured syslog server. Syslog uses TES (RFC 5425) when available, with support for mutual TLS (client certificate) authentication so that the syslog server can verify the isolator's identity. The syslog server address and port are configurable via the administrative interface and the syslog server is automatically added to the system whitelist for isolator-originated traffic only; the syslog server address is NOT automatically whitelisted for camera traffic.

[0131] The network isolator also supports email notifications for critical events, including tamper detection, firmware update availability, and repeated blocked access attempts from the same source,via SMTP with STARTTLS. The SMTP server is automatically added to the system whitelist for isolator-originated traffic.Physical Security, Hardware Protections, and Reset Mechanism

[0132] The network isolator provides a graduated physical reset button mechanism with LED feedback that enables field technicians to perform different levels of device reset without counting seconds. The reset button supports four distinct press durations, each triggering a different recovery action: (1) Short press (less than 3 seconds) — resets the administrator password to the factory default while preserving all other configuration including whitelist entries and network settings; (2) Medium press (3 to 9 seconds) — re-enables the administrative interface if it has been disabled and resets the administrator password to factory default, while preserving all other configuration; (3) Long press (10 to 19 seconds) — full factory reset restoring all settings to factory defaults; (4) Very long press (20 or more seconds) — PURGE mode. During a press, the LED blink rate changes at each threshold while the button is held, providing visual confirmation of the selected action level without requiring the user to count seconds. This LED-mirrored threshold feedback is a novel improvement over conventional timed-hold reset mechanisms.

[0133] PURGE mode, triggered by a very long press, executes a thorough destruction of all sensitive device credentials: (a) the mutable protected-flash trust material (per-device HMAC for cloud authentication, configuration encryption key, TOTP secrets, TLS private keys) is destroyed by overwriting the flash sectors containing the wrapping key with random data from the hardware TRNG; (b) the overwritten sectors are read back and verified to contain the random data pattern before PURGE is declared successful; (c) device-indelible identity material in OTP fuses (serial, hardware unique key seed, anti-rollback counter base, signing CA fingerprint) is preserved to enable post-incident forensic device identification; (d) on successful completion, a confirmation message is emitted on the serial console. PURGE does not require a lengthy flash sector overwrite because the cryptographic master key is derived from the OTP material and a mutable nonce in protected flash; destroying the nonce cryptographically shreds all derived key material instantly.

[0134] If a physical tamper detection switch is present on the network isolator hardware, a tamper event triggers: (a) immediate logging of the tamper event to the security / forensic log and, if syslog is configured, to the syslog server; (b) optionally (configurable), wiping all stored credentials and reverting the device to first-login state. If credential wiping is configured, the tamper log entry is written before the wipe occurs. The firmware supports the tamper switch via a GPIO pin whose assign

[0135] The MPU is configured to isolate sensitive memory regions — including cryptographic key material, password hashes, and TOTP secrets — from general firmware access and to mark key material memory as non-executable. The MPU configuration is verified to persist across STM32H5 low-power modes (Stop and Standby modes) and is re-armed before any user task is permitted toexecute after wakeup from low-power mode. Additionally, FreeRTOS heap randomization is implemented at boot using the hardware TRNG to select the heap base address offset within a larger-than-required heap region, constrained to MPU-compatible alignment boundaries (for example, 32-byte granularity), providing approximately five bits of entropy to defeat template-based exploitation attacks that rely on fixed heap addresses.

[0136] An optional administrative interface lockdown feature allows the administrator to completely disable the administrative web interface (port 3854) once the device is fully configured and operational. When the interface is disabled, it can only be re-enabled via a physical medium press of the reset button. This hardens the device against remote administrative interface attacks in environments where the device configuration is stable and remote management access is not required.International Security Certification Compliance

[0137] The network isolator is designed for compliance with the following international security certification standards applicable to government and enterprise security products: FIPS 140-3 Level 3, PSA Certified Level 3, and SESIP (Security Evaluation Standard for loT Platforms) Level 3. These standards require, respectively: FIPS 140-3 Level 3 — validated cryptographic module with physical tamper detection and response, role-based authentication, and the use of FIPS-approved algorithms; PSA Certified Level 3 — hardware-based root of trust, secure boot, firmware update with cryptographic verification, isolation between security domains, and resistance to hardware and software attacks; SESIP Level 3 — lifecycle protection, secure storage of secrets, authenticated firmware updates, and security certification independently verified by an accredited testing laboratory.

[0138] To comply with these standards, the network isolator employs exclusively FIPS-approved cryptographic algorithms: AES-256-GCM for symmetric encryption; SHA-256 and SHA-384 for hashing; ECDSA P-256 for digital signatures; ECDHE P-256 for key exchange; HMAC-SHA-256 for message authentication; PBKDF2 -HMAC-SHA-256 with a minimum of 100,000 iterations for passphrase-based key derivation; and Ed25519 for firmware signing. All cryptographic random number generation uses the STM32H5 hardware TRNG, validated per NIST SP 800-90B. The use of these algorithms via the hardware cryptographic accelerators ensures that the cryptographic implementation can achieve FIPS 140-3 validation.

[0139] An independent accredited testing laboratory must perform security evaluation and testing of the network isolator before certification claims can be formally made. The design described herein is intended to satisfy the technical requirements of the stated certification levels; however, formal certification is obtained only after successful completion of the evaluation process. Government and enterprise customers are advised to obtain confirmation of the current certification status of the specific firmware version deployed.

[0140] Alternative compliance embodiments may target additional or alternative certification schemes applicable to specific markets, including: Common Criteria (ISO / IEC 15408) at EAL4+ or higher; IEC 62443 Level 3 for industrial cybersecurity; DISA STIG (Security Technical Implementation Guide) compliance for U.S. Department of Defense deployments; and national security certification schemes such as France's ANSSI Certification de Securite de Premier Niveau (CSPN) or Germany's BSI Zertifizierung.Runtime Isolation Self-Test and Fail-Safe Operation

[0141] A fundamental safety invariant of the network isolator is that no packet may traverse the isolation boundary from CamPort to WanPort (or vice versa) without passing through the whitelist enforcement engine and NAT state machine. To continuously validate this invariant at runtime, the network isolator implements a periodic non-IP marker frame cross-port isolation self-test. Every 60 minutes, the network isolation self-test module injects a marker Ethernet frame using a reserved EtherType outside the IP family on one port and verifies that the frame does not appear on the other port. Because the marker frame uses a non-IP EtherType, it is invisible to and not processed by the NAT engine; a dedicated test receiver hooks the raw ingress of the opposite port to detect the marker. If the marker frame is detected on the wrong port — indicating a firmware defect, buffer aliasing error, or hardware fault that would allow packet leakage between ports — the device immediately transitions to safe mode, in which both Ethernet ports are taken down, all traffic is dropped, the red LED is illuminated solid, and an alert is forwarded to the configured syslog server. This self-test directly validates the product's core security guarantee at runtime rather than relying solely on pre-deployment testing.

[0142] At boot, the isolator additionally verifies that the CamPort and WanPort are correctly configured as separate MAC / IP stacks with no shared frame buffers. A firmware-level audit verifies the Ethernet controller configuration registers to confirm physical-layer separation before any network traffic is accepted on either port.Configuration Management and Multi-Device Deployment

[0143] The network isolator supports export of the device configuration — including all whitelist entries, network settings, authentication configuration, and log settings — as an encrypted archive file. The configuration export is encrypted using AES-256-GCM with a key derived from a user-supplied passphrase using PBKDF2-HMAC-SHA-256 with a minimum of 100,000 iterations. The export file includes an HMAC for integrity verification. By default, the export excludes secrets (password hashes, TOTP seeds, TLS private keys, LDAP bind credentials). An "include secrets" export option is available but requires re-authentication and displays a security warning. The exported configuration file can be imported into another network isolator of the same firmware version to replicate the configuration, facilitating consistent deployment across fleets of devices protectingmultiple cameras.

[0144] The administrative interface provides a bulk whitelist input mechanism supporting paste of multiple entries in the format: IP[ / CIDR][:port[,port,...]][@MAC][#label], with one entry per line. A CSV file upload alternative is also supported. After completing the camera setup wizard, the interface prompts the administrator to export a configuration backup and displays a persistent (dismissible) banner on the dashboard if no backup has been made.

[0145] The administrative interface provides support for 802. IX port-based network access control on the WanPort, for government and enterprise network deployments that require the isolator to authenticate itself to the network switch before being granted LAN access. The EAP-TLS (certificatebased) authentication method is supported. When 802. IX is enabled, the DHCP and link-local address assignment state machine is deferred until 802. IX authentication succeeds or the 802. IX timeout (default 30 seconds) expires.Summary of Novel Technical Innovations

[0146] The following technical innovations embodied in the network isolator are identified as potentially novel combinations or applications that distinguish the present invention from prior art network firewalls, NAT devices, and loT security gateways. These innovations are described throughout this specification and are reflected in the claims appended hereto.Camera Discovery and Binding Innovations

[0147] The atomic first- wins multi-method single-camera discovery system concurrently runs three discovery pathways — embedded DHCP server, passive ARP observation, and manual administrative entry — against a single isolated subnet and commits on the first successful discovery method using a compare-and-swap atomic operation, cancelling all remaining methods in the same atomic step. This eliminates race conditions between discovery methods and enables the device to work with DHCP-capable cameras, static-IP cameras, and operator-provisioned cameras without requiring the installer to configure the discovery method.

[0148] The adaptive CamPort self-IP assignment mechanism discovers the camera's subnet and automatically assigns the CamPort gateway IP address by probing the subnet's ".101" host first (with ARP safety), falling back to ".254" on conflict, and refusing to assign any address if a third host is detected on the subnet. This mechanism enforces the single-camera subnet invariant as a runtime check while automatically configuring the gateway address for any camera subnet.

[0149] The combination of ARP-binding lock with link-cycle re-validation provides two complementary protections: it rejects any ARP packet that would change the bound camera MAC-to-IP mapping (defeating ARP spoofing on the isolated subnet) and automatically re-runs full discovery when the CamPort Ethernet link cycles (defeating physical camera replacement attacks). The additional CamPort source-MAC pinning operates at Layer 2, dropping frames whose sourceMAC does not match the bound camera MAC, providing defense in depth below the IP layer. Learning Mode and Setup Innovations

[0150] The non-forwarding learning mode with active DNS resolution captures all camera-originated traffic as whitelist candidates without forwarding any of it to the WAN, while maintaining an active DNS proxy so that hostname context (resolved domain names) is captured alongside destination IP addresses and ports. The power-loss-safe transaction log for learning mode state ensures that partial captures are not lost. The combination of these features enables non-technical installers to make informed whitelist decisions without network expertise.

[0151] The hybrid camera-brand detection system combining MAC OUI lookup, DNS query pattern matching, and destination IP range matching against a static firmware-resident vendor preset database enables pre-filled whitelist suggestions for known camera brands without requiring a network-accessible vendor catalog, and without an unbounded database that would require regular updates.Runtime Safety and Integrity Innovations

[0152] The periodic non-IP marker-frame cross-port isolation self-test continuously validates the product's core safety invariant — that no packet leaks between ports without whitelist authorization — at runtime. The use of a non-IP EtherType marker frame that is invisible to the NAT engine but detectable by a dedicated raw-ingress receiver enables this self-test without affecting normal traffic. Any failure transitions the device to a fail-shut state in which both ports are taken down.

[0153] The fail-shut secure boot mechanism, in which both Ethernet ports remain inactive if both primary and backup firmware partitions fail signature verification, ensures that a compromised device cannot masquerade as a functional security boundary. The two-plane traffic segregation architecture, distinguishing forwarding-plane traffic (subject to the operator's whitelist and NAT) from control-plane traffic (the device's own services), with a hardcoded egress IP-and-port allowlist for device-originated traffic, prevents compromised firmware from exfiltrating data via the device's own legitimate services.Cryptographic Key Management Innovations

[0154] The split root-of-trust architecture separates manufacturing-time indelible identity anchors in OTP fuses (serial number, hardware unique key seed, anti-rollback counter base, signing CA fingerprint) from mutable trust material in MPU-protected, AES-GCM-wrapped flash (per-device HMAC for cloud authentication, monotonic firmware counter, configuration encryption key). The PURGE operation destroys only the mutable flash material (by destroying the wrapping key nonce) while preserving the OTP-anchored identity for post-incident forensics.

[0155] The PURGE operation with verified readback writes random TRNG data to the destroyed flash sectors and reads them back to verify the overwrite pattern before declaring PURGE successful,providing runtime confirmation of secure destruction. The anti-rollback mechanism anchoring the firmware monotonic version counter in OTP fuses and including it in every update-check request — with server-side rejection of requests where the counter does not exceed the last seen value — defeats firmware downgrade attacks.Logging and Operator Experience Innovations

[0156] The double-buffered, chained-HMAC log with atomic pointer swap provides tamper-evident, power-loss-safe log appends on a device with no battery-backed RAM. The combination of a write-once forensic sector (not erasable by factory reset) with the FIFO operational logs provides both tamper-evident operational logging and indelible forensic record-keeping within a single embedded device. The click-through blocked-log to whitelist dialog, in which clicking a blocked traffic entry opens a pre-filled whitelist addition dialog, turns security telemetry into security policy configuration with a single user gesture.

[0157] The graduated reset button with LED-mirrored threshold feedback, in which the LED blink rate changes at each press-duration threshold while the button is held, enables field technicians to perform four distinct recovery operations (password reset, admin UI re-enable, factory reset, PURGE) without counting seconds.Forwarding Transparency and Out-of-Band Administration Innovations

[0158] The dual-plane transparent architecture simultaneously operates two completely independent traffic planes — a forwarding plane and a control plane — that share a single WAN IP address, enabling the network isolator to remain fully transparent and undetectable in the packetforwarding path while maintaining an independently accessible administrative interface that is invisible to the peripheral accessory. This architecture is novel in that it achieves out-of-band administrative access without requiring a dedicated physical management port or separate management network: the administrative interface is logically out-of-band — inaccessible from the peripheral-accessory side and excluded from all NAT forwarding rules — while remaining physically in-band on the same WAN port as all forwarded traffic.

[0159] In a first embodiment of the multiplexed WAN port, traffic discrimination is achieved by transport-layer port number: packets destined for the reserved administrative port are assigned to the management function, and all other packets are evaluated for pass-through forwarding. In a second embodiment, the second input / output port is configured to respond to two distinct IP addresses — a management IP address and a forwarding IP address — such that the management function and the pass-through forwarding function are separated at the IP layer rather than the port layer. In a third embodiment, both IP address discrimination and port discrimination are applied in combination, providing redundant multiplexing that unambiguously assigns every arriving packet to exactly one function. These embodiments are not mutually exclusive, and further multiplexing methods operableon the second input / output port are contemplated within the scope of the invention. In all embodiments, the peripheral accessory attached to the first input / output port has no network path to the management function by any method.

[0160] The port-discriminating traffic classifier assigns each incoming WAN-side connection to either the forwarding plane or the control plane based on its destination port number and source allow-list status, implementing the following novel combination: connections to the reserved administrative port are unconditionally routed to the control plane regardless of source allow-list status; connections to permitted forwarding ports from whitelisted sources are transparently proxied to the peripheral accessory via NAT; and connections to non-administrative ports from non-whitelisted sources receive HTTPS redirects to the administrative interface, converting connection failures into guided administrative-access paths. This three-branch discrimination algorithm enables a single WAN IP address to simultaneously serve as the transparent proxy address of the peripheral accessory, the gateway to the administrative interface, and the entry point for initial installer discovery — all without exposing the existence or location of the security device to the peripheral accessory.

[0161] The control-plane-only mDNS advertisement generates an mDNS service announcement for the administrative interface exclusively on the WAN port, using a device-specific hostname derived from the WAN MAC address, and ensures that this advertisement is never forwarded to or detectable from the peripheral-accessory side of the isolation boundary. This means that installers can discover the administrative interface via zero-configuration network browsing from the WAN side, while a compromised peripheral accessory on the CamPort side cannot detect the mDNS announcement, cannot resolve the administrative hostname, and cannot infer from network traffic analysis that a security device is present on the network.

[0162] The NAT transparency security property — the mutual invisibility of the network isolator to both the peripheral accessory and the computer network — is a distinct security benefit beyond the traffic filtering functions. By ensuring that neither side can distinguish the isolator from an ordinary router or direct connection, the architecture defeats topological reconnaissance by compromised peripherals and prevents adaptive attacks that rely on identifying the make, model, or capabilities of an intermediate security device. This property is enforced by the NAT engine's address rewriting, which makes the camera appear to be directly reachable at the WAN IP address from the network side, and by the isolator presenting itself as a plain default gateway to the camera with no distinguishing management surface on any CamPort-reachable address or port.

[0163] It is contemplated that any embodiment discussed in this specification can be implemented with respect to any method of the invention, and vice versa. It will also be understood that particular embodiments described herein are shown by way of illustration and not as limitations of theinvention. The principal features of this invention can be employed in various embodiments without departing from the scope of the invention. Those skilled in the art will recognize, or be able to ascertain using no more than routine experimentation, numerous equivalents to the specific procedures described herein. Such equivalents are considered to be within the scope of this invention and are covered by the claims.

[0164] All publications and patent applications mentioned in the specification are indicative of the level of skill of those skilled in the art to which this invention pertains. All publications and patent applications are herein incorporated by reference to the same extent as if each individual publication or patent application was specifically and individually indicated to be incorporated by reference. Incorporation by reference is limited such that no subject matter is incorporated that is contrary to the explicit disclosure herein, no claims included in the documents are incorporated by reference herein, and any definitions provided in the documents are not incorporated by reference herein unless expressly included herein.

[0165] The use of the word “a” or “an” when used in conjunction with the term “comprising” in the claims and / or the specification may mean “one,” but it is also consistent with the meaning of “one or more,” “at least one,” and “one or more than one.” The use of the term “or” in the claims is used to mean “and / or” unless explicitly indicated to refer to alternatives only or the alternatives are mutually exclusive, although the disclosure supports a definition that refers to only alternatives and “and / or.” Throughout this application, the term “about” is used to indicate that a value includes the inherent variation of error for the device, the method being employed to determine the value, or the variation that exists among the study subjects.

[0166] As used in this specification and claim(s), the words “comprising” (and any form of comprising, such as “comprise” and “comprises”), “having” (and any form of having, such as “have” and “has”), “including” (and any form of including, such as “includes” and “include”) or “containing” (and any form of containing, such as “contains” and “contain”) are inclusive or open-ended and do not exclude additional, unrecited elements or method steps. In embodiments of any of the compositions and methods provided herein, “comprising” may be replaced with “consisting essentially of’ or “consisting of’. As used herein, the phrase “consisting essentially of’ requires the specified integer(s) or steps as well as those that do not materially affect the character or function of the claimed invention. As used herein, the term “consisting” is used to indicate the presence of the recited integer (e.g., a feature, an element, a characteristic, a property, a method / process step or a limitation) or group of integers (e.g., feature(s), element(s), characteristic(s), propertie(s), method / process steps or limitation(s)) only.

[0167] The term “or combinations thereof’ as used herein refers to all permutations and combinations of the listed items preceding the term. For example, “A, B, C, or combinations thereof’is intended to include at least one of: A, B, C, AB, AC, BC, or ABC, and if order is important in a particular context, also BA, CA, CB, CBA, BCA, ACB, BAC, or CAB. Continuing with this example, expressly included are combinations that contain repeats of one or more item or term, such as BB, AAA, AB, BBC, AAABCCCC, CBBAAA, CAB ABB, and so forth. The skilled artisan will understand that typically there is no limit on the number of items or terms in any combination, unless otherwise apparent from the context.

[0168] As used herein, words of approximation such as, without limitation, “about”, "substantial" or "substantially" refers to a condition that when so modified is understood to not necessarily be absolute or perfect but would be considered close enough to those of ordinary skill in the art to warrant designating the condition as being present. The extent to which the description may vary will depend on how great a change can be instituted and still have one of ordinary skilled in the art recognize the modified feature as still having the required characteristics and capabilities of the unmodified feature. In general, but subject to the preceding discussion, a numerical value herein that is modified by a word of approximation such as “about” may vary from the stated value by at least ±1, 2, 3, 4, 5, 6, 7, 10, 12, 15, 20 or 25%.

[0169] All of the devices and / or methods disclosed and claimed herein can be made and executed without undue experimentation in light of the present disclosure. While the devices and methods of this invention have been described in terms of preferred embodiments, it will be apparent to those of skill in the art that variations may be applied to the devices and / or methods and in the steps or in the sequence of steps of the method described herein without departing from the concept, spirit and scope of the invention. All such similar substitutes and modifications apparent to those skilled in the art are deemed to be within the spirit, scope and concept of the invention as defined by the appended claims.

Claims

WHAT IS CLAIMED IS:

1. A method for improving computer network security comprising:providing a network isolator with a first input / output port operatively connected to a single peripheral accessory on an isolated peripheral- side subnet and a second input / output port operatively connected to a computer network, wherein the first input / output port is physically separated from the second input / output port;enforcing a single-device invariant on the peripheral-side subnet by detecting whether more than one network-addressable device is present on the subnet connected to the first input / output port and blocking all traffic forwarding if more than one such device is detected;binding a discovered network address and hardware address of the peripheral accessory to the first input / output port and rejecting any address-resolution protocol packet on the first input / output port that would alter the bound address association; andevaluating every packet received on the first input / output port and every packet received on the second input / output port against an internal allow-list database and forwarding each packet across the isolation boundary only if that packet's attributes match a permitted entry in the allow-list database, and blocking and logging all non-matching packets;wherein the allow-list database specifies permitted traffic by one or more of: source IP address, destination IP address, IP address range or subnet, transport-layer port number, network protocol, time of day, or any combination thereof; andwherein the permit or block decision is made independently of whether the packet was initiated by the peripheral accessory or by a node on the computer network.

2. The method for improving computer network security according to claim 1, wherein the network isolator is operatively connected to a plurality of video cameras via a plurality of individual first input / output ports, and wherein operating the network isolator comprises individual communications with each video camera via the corresponding plurality of individual first input / output ports.

3. The method according to claim 2, further comprising evaluating packets originating from the computer network against the internal allow-list database and forwarding to corresponding individual video cameras only those packets whose attributes match a permitted entry in the allow-list database.

4. A network isolator for improving computer network security when using a peripheral accessory,the network isolator comprising:a first input / output port configured to be operatively connected to a peripheral accessory on an isolated peripheral-side subnet;a second input / output port configured to be operatively connected to a computer network, wherein the first input / output port is physically separated from the second input / output port; and a controller configured to:enforce a single-device invariant on the peripheral-side subnet by detecting the presence of more than one network-addressable device on the subnet connected to the first input / output port and blocking all traffic forwarding if more than one such device is detected;bind a discovered network address and hardware address of the peripheral accessory to the first input / output port and reject any address-resolution protocol packet received on the first input / output port that would alter the bound address association;drop any packet received on the first input / output port whose source hardware address does not match the bound hardware address of the peripheral accessory;re-initiate peripheral accessory discovery upon detection of a link-down followed by a link-up transition on the first input / output port;evaluate any packet received on the first input / output port or the second input / output port against an internal allow-list database and forward the packet across the isolation boundary only if the packet's attributes match a permitted entry in the allow-list database; andblock and log any packet whose attributes do not match any permitted entry in the allow-list database;wherein the allow-list database specifies permitted traffic by one or more of: source IP address, destination IP address, IP address range or subnet, transport-layer port number, network protocol, time of day, or any combination thereof, and wherein the permit or block decision is made independently of whether the packet was initiated by the peripheral accessory or by a node on the computer network.

5. A network isolator for containing network communications of a directly- attached untrusted peripheral accessory, comprising:a peripheral-side port forming a dedicated isolated subnet to which exactly one peripheral accessory is directly connected;a network-side port connected to a trusted computer network;a single-device enforcement module configured to verify that no more than one network-addressable device is present on the isolated subnet and to block all traffic forwarding if the singledevice invariant is violated;a device-binding module configured to associate the peripheral accessory's hardware address with its network address upon discovery and to reject any packet on the peripheral-side port whose source hardware address does not match the associated hardware address;a policy enforcement module configured to evaluate all packets traversing in either direction against an internal allow-list database and to forward only packets whose source address, destination address, port, protocol, and any configured time constraint collectively match a permitted database entry; anda content sanitization module configured to inspect the payload of permitted packets and remove embedded content that is inconsistent with the protocol type or content format of the matched allow-list entry before forwarding;wherein the network isolator is architecturally distinct from a perimeter firewall in that the threat being contained originates from a device physically attached to the peripheral-side port, and wherein the single-device invariant, device-binding, link-cycle re-validation, and content sanitization functions collectively enforce containment of that device rather than filtering of perimeter traffic.

6. A method of operating a network security device as a transparent network intermediary while maintaining an out-of-band administrative interface, comprising:operating, by a network security device interposed between a peripheral accessory and a computer network, a forwarding plane that performs network address translation between a first input / output port connected to the peripheral accessory and a second input / output port connected to the computer network, such that the peripheral accessory is represented on the computer network by the network security device's own network address and the network security device does not appear as a distinct hop in the forwarding path;operating, by the network security device simultaneously with the forwarding plane, a control plane comprising an administrative interface accessible on the network security device's network address at a reserved port number that is explicitly excluded from the network address translation forwarding rules, such that a node on the computer network connecting to the reserved port number is served the administrative interface of the network security device and is not forwardedto the peripheral accessory;preventing all traffic originating from the peripheral accessory from reaching the control plane by excluding the reserved port from all allow-list entries applicable to the first input / output port; and maintaining complete isolation between the forwarding plane and the control plane such that the peripheral accessory has no network path, address, or port by which it can access or interact with the administrative interface.

7. The method of operating a network security device, as in claim 6, further comprising advertising the administrative interface via mDNS on the second input / output port under a device-specific hostname, wherein the mDNS advertisement is generated by the control plane and is not forwarded to the peripheral accessory via the first input / output port.

8. The method of operating a network security device, as in claim 6, further comprising, upon receiving a packet on the second input / output port destined for the network security device's network address, determining whether the destination port matches the reserved administrative port number, and if so terminating the connection at the network security device's administrative interface without forwarding the packet to the peripheral accessory, and if not evaluating the packet against the allow-list database and forwarding it to the peripheral accessory only if it matches a permitted entry.

9. The method of operating a network security device, as in claim 6, further comprising selectively disabling the administrative interface such that the reserved port is no longer served on the network security device's network address, after which the network security device presents no administrative surface to the computer network and the network address serves exclusively as the forwarding-plane address of the peripheral accessory.

10. A network isolator comprising:a first input / output port configured to be operatively connected to a peripheral accessory;a second input / output port configured to be operatively connected to a computer network, wherein the first input / output port is physically separated from the second input / output port; and a controller configured to multiplex the second input / output port to simultaneously serve two distinct functions using a single network address assigned to the second input / output port:(i) a pass-through forwarding function, in which traffic addressed to the network address of the second input / output port on permitted forwarding ports is transparently forwarded to the peripheral accessory via the first input / output port; and(ii) a management function, in which traffic addressed to the same network address of the second input / output port is terminated at the controller and serves as the administrative interface of the network isolator;wherein the controller comprises a traffic classifier configured to assign each packet arriving on the second input / output port to either the pass-through forwarding function or the management function, andwherein the peripheral accessory has no network path, address, or port by which it can access or interact with the management function.

11. The network isolator of claim 10, wherein the traffic classifier assigns a packet to the management function if the packet's destination transport-layer port number matches a reserved administrative port number, and assigns a packet to the pass-through forwarding function if the destination port does not match the reserved administrative port number and the packet's attributes match a permitted entry in an allow-list database,wherein the reserved administrative port number is excluded from all allow-list entries governing pass-through forwarding, andwherein the first use of the network address of the second input / output port for management may precede configuration of pass-through forwarding entries, such that the management function is accessible for initial device configuration before any forwarding policy is established.

12. The network isolator of claim 10, wherein the second input / output port is configured to respond to two distinct IP addresses:a first IP address exclusively associated with the management function, to which packets are terminated at the controller and served as the administrative interface of the network isolator; and a second IP address exclusively associated with the pass-through forwarding function, to which packets are transparently forwarded by the controller to the peripheral accessory via the first input / output port;wherein the peripheral accessory is not addressable at the first IP address from the first input / output port side, and the management function is not reachable via the second IP address.

13. The network isolator of claim 10, wherein the traffic classifier applies a multi-dimensional multiplexing algorithm comprising:(a) a destination IP address filter that assigns traffic addressed to a designated management IP address to the management function; and(b) a destination transport-layer port filter that, for traffic addressed to a forwarding IP address, assigns the packet to the management function if the destination port matches a reserved administrative port number, and to the pass-through forwarding function if the destination port matches a permitted forwarding port in an allow-list database;wherein both the IP address filter and the port filter may be applied in combination to unambiguously assign any arriving packet to exactly one of the two functions.

14. The network isolator of claim 10, wherein the network isolator is architecturally distinct from a conventional network firewall in that:(a) on the network isolator, the single network address of the second input / output port serves both as the management address of the network isolator and as the forwarding-plane address transparently representing the peripheral accessory to the computer network, with a single physical port multiplexed to serve both purposes; and(b) on a conventional network firewall, the WAN-side address serves exclusively as the management address or external routing address of the firewall, and the firewall does not represent any directly-attached device transparently at that same address;wherein the multiplexing of management and pass-through functions at a single physical port and single IP address is enforced by the traffic classifier and is a distinct security and deployment advantage requiring no dedicated management network or separate physical management port.

15. A network security device comprising:a first network port connected to a peripheral-side subnet on which a peripheral accessory is directly attached;a second network port connected to a computer network;a forwarding plane comprising a network address translation engine configured to relay permitted traffic between the first network port and the second network port such that the peripheral accessory is represented on the computer network by the network security device's own network address, and such that nodes on the computer network communicating with permitted ports are transparently connected to the peripheral accessory without detection of the network security device as an intermediate hop;a control plane comprising an administrative interface server configured to serve an administrative interface to nodes connecting to the network security device's own network address at a reserved port number; anda port-discriminating traffic classifier configured to, for each packet received on the second network port: if the destination port matches the reserved port number, terminate the connection at the administrative interface server without evaluation by the forwarding plane; and if the destination port does not match the reserved port number, evaluate the packet against an allow-list database and forward to the peripheral accessory only if permitted;wherein the reserved port number is excluded from all allow-list entries governing traffic on the first network port, such that the peripheral accessory has no network path to the administrative interface server.

16. A method for improving computer network security comprising the following steps: providing a network isolator with a first input / output port operatively connected to a peripheral accessory and a second input / output port operatively connected to the computer network, wherein the first input / output port is physically separated from the second input / output port; operating the network isolator to communicate with the peripheral accessory via the first input / output port by passing a command from the computer network received via the second input / output port; andoperating the network isolator to communicate with the computer network via the second input / output port by evaluating traffic originating from the peripheral accessory against an internal allow-list database and forwarding traffic that matches a permitted entry in the allow- list database to the computer network, and by evaluating traffic originating from the computer network against the internal allow-list database and forwarding traffic that matches a permitted entry in the allow-list database to the peripheral accessory;wherein the network isolator is configured to permit or block any packet traversing in either direction between the first input / output port and the second input / output port based solely on whether that packet's attributes match a permitted entry in an internal allow-list database maintained by the network isolator, regardless of whether the packet was initiated by the peripheral accessory or by a node on the computer network; andwherein the internal allow-list database specifies permitted traffic by one or more of: source IP address, destination IP address, IP address range or subnet, transport-layer port number, network protocol, time of day, or any combination thereof.

17. The method for improving computer network security according to claim 16, wherein the peripheral accessory is a video camera and the signal from the peripheral accessory is a video signal.

18. The method for improving computer network security according to claim 16, wherein the network isolator is operatively connected to a plurality of video cameras via a plurality of individual first input / output ports, and wherein operating the network isolator comprises individual communications with each video camera via the corresponding plurality of individual first input / output ports.

19. The method for improving computer network security according to claim 18, further comprising passing individual commands from the computer network to corresponding individual video cameras.

20. The method for improving computer network security according to claim 18, further comprising evaluating video signals from the plurality of video cameras received via the corresponding plurality of individual first input / output ports against the internal allow-list database and forwarding those signals that match a permitted entry to the computer network, regardless of whether the signals were initiated by the cameras autonomously or in response to traffic from the computer network.

21. The method for improving computer network security according to claim 20, further comprising passing individual video signals from the plurality of video cameras to one or more computers on the computer network, each of the one or more computers designated to receive one or more individual video signals from corresponding video cameras.

22. The method for improving computer network security according to claim 17, wherein operating the network isolator comprises passing the video signal from the video camera received via the first input / output port to one or more computers on the computer network designated to receive the video signal from the video camera.

23. The method for improving computer network security according to claim 16, further comprising operating the network isolator to inspect the signal generated by the peripheral accessory and remove unauthorized data and commands therefrom.

24. The method for improving computer network security according to claim 16, wherein the network isolator is operatively connected via the first input / output port to one or more peripheral accessories via a first connection network, and wherein the network isolator is operatively connected via the second input / output port to the computer network via a second connection network, such that the one or more peripheral accessories and the first connection network are separated from the computer network and from the second connection network by the network isolator.

25. A network isolator for improving computer network security when using a peripheral accessory, the network isolator comprising:a first input / output port configured to be operatively connected to the peripheral accessory; a second input / output port configured to be operatively connected to the computer network, wherein the first input / output port is physically separated from the second input / output port; anda controller configured to:communicate with the peripheral accessory via the first input / output port by passing a command from the computer network received via the second input / output port; evaluate any packet received on the first input / output port or the second input / output port against an internal allow-list database and forward the packet across the isolation boundary only if the packet's attributes match a permitted entry in the allow-list database; block any packet received on either the first input / output port or the second input / output port whose attributes do not match any permitted entry in the allow-list database, and log the blocked packet;wherein the allow-list database specifies permitted traffic by one or more of: source IP address, destination IP address, IP address range or subnet, transport-layer port number, network protocol, time of day, or any combination thereof, and wherein the permit or block decision is made independently of whether the packet was initiated by the peripheral accessory or by a node on the computer network.

26. The network isolator according to claim 25, wherein the first input / output port comprises a plurality of individual first input / output ports, each individual first input / output port configured for operative connection to a designated peripheral accessory.

27. The network isolator according to claim 26, wherein at least one individual first input / output port is a virtual first port configured for designated wireless communication with a corresponding peripheral accessory.

28. The network isolator according to claim 26, wherein at least one individual first input / output port is a physical connector configured for designated wired attachment and communication with a corresponding peripheral accessory.

29. The network isolator according to claim 25, wherein the second input / output port comprises a plurality of individual second input / output ports, each individual second input / output portconfigured for operative connection to a designated computer on the computer network.

30. The network isolator according to claim 26, wherein at least one individual second input / output port is a virtual second port configured for designated wireless communication with a corresponding computer on the computer network.

31. The network isolator according to claim 26, wherein at least one individual second input / output port is a physical connector configured for designated wired attachment and communication with a corresponding computer on the computer network.

32. The network isolator according to claim 25, wherein the controller is further configured to inspect packets permitted by the allow-list database and, prior to forwarding, remove any data, metadata, or embedded commands that are not consistent with the declared protocol type or content format for that traffic class, regardless of the direction in which the packet is travelling to the computer on the computer network designated to receive the signal.

33. A method for discovering and binding a peripheral accessory to a network gateway device comprising:concurrently initiating, on a gateway device comprising a first network port isolated from a second network port, a plurality of discovery procedures on the first network port, the plurality of procedures comprising at least an embedded DHCP server procedure, a passive ARP observation procedure, and a manual operator input procedure;atomically committing, upon the first of the plurality of discovery procedures to successfully identify exactly one device on the first network port, the discovered device address as the bound peripheral accessory address; andcancelling all remaining uncommitted discovery procedures in the same atomic operation in which the commit occurs; wherein no other device address is accepted for the first network port after the atomic commit.

34. The method according to claim 33, further comprising: upon determining that the peripheral accessory has a statically-configured IP address via the passive ARP observation procedure, selecting a gateway IP address for the first network port by issuing an ARP probe to a first candidate address on the discovered camera subnet; assigning the first candidate address if no ARP response is received; and falling back to a second candidate address and issuing an ARP probe thereto if the first candidate address is in use, wherein if any third IP address on the discovered subnet responds to an ARP scan, generating an error condition indicating that more than one non-gateway device is present on the isolated subnet.

35. The method according to claim 33, further comprising: upon completion of device discovery, binding the discovered MAC address and IP address of the peripheral accessory; dropping any ARP packet on the first network port that would modify the bound MAC-to-IP association; monitoring the first network port for link-state transitions; and upon detection of a link-down followed by a link-up transition on the first network port, automatically re-initiating the plurality of discovery procedures to verify that the same device is connected.

36. A method for configuring access control policy for an isolated network peripheral comprising:operating a gateway device in a non-forwarding learning mode for a configurable time period in which all traffic originating from a peripheral accessory on a first isolated network port is captured and recorded as candidate access control entries but not forwarded to a second network port;during the non-forwarding learning mode, operating a DNS proxy on the first network port that resolves DNS queries from the peripheral accessory and records the correspondence between resolved hostnames and destination IP addresses and port numbers;upon completion of the non-forwarding learning mode, presenting to an operator a list of the captured candidate destinations including resolved hostnames associated with each destination IP address; and creating access control whitelist entries based on operator selection from the presented list.

37. The method according to claim 36, further comprising identifying a vendor of the peripheral accessory by a combination of: matching the peripheral accessory's MAC address against a database of vendor-specific Organizationally Unique Identifiers; matching DNS queries observed during the non-forwarding learning mode against a database of vendor-specific domain name patterns; and matching destination IP addresses observed during the non-forwarding learning mode against a database of vendor-specific IP address ranges; and pre-populating the candidate access control entries with entries from a vendor-specific preset template corresponding to the identified vendor.

38. A network gateway device comprising:a first network port on an isolated subnet connected to a peripheral accessory;a second network port connected to a computer network;a packet forwarding engine configured to enforce a whitelist-based access control policy between the first and second network ports;a cross-port isolation self-test module configured to: periodically inject a marker Ethernet frame using a reserved EtherType on the first network port; monitor the second network port for receipt of the marker frame; and upon detecting the marker frame on the second network port, transition the gateway device to a fail-shut state in which no traffic is forwarded on either network port; wherein the marker Ethernet frame is not processed by the packet forwarding engine due to its reserved non-IP EtherType.

39. The network gateway device according to claim 38, further comprising a secure boot subsystem configured to: verify a cryptographic signature of a primary firmware partition at boot; verify a cryptographic signature of a backup firmware partition if the primary verification fails; and refrain from activating either the first network port or the second network port if both the primary and backup firmware partition signature verifications fail.

40. The network gateway device according to claim 38, wherein the packet forwarding engine is configured as a first software task and an administrative interface is configured as a second software task, wherein the first and second software tasks communicate exclusively through message queues with no shared memory regions, and wherein a hardware memory protection unit enforces the memory isolation between the first and second software tasks.

41. A network gateway device comprising:a hardware root-of-trust comprising OTP fuses storing immutable identity anchors including a device serial number, a hardware unique key seed, an anti-rollback counter base value, and a signing CA fingerprint;a protected flash memory region storing mutable trust material including a per-device HMAC key for server authentication, a configuration encryption key, and a monotonic firmware version counter, wherein the mutable trust material is encrypted using a key derived from the OTP-anchored hardware unique key seed; anda PURGE mechanism configured to destroy the mutable trust material by overwriting the protected flash memory region with data from a hardware true random number generator, reading back the overwritten memory region, and verifying that the read-back data matches the written random data pattern before reporting PURGE success;wherein the OTP-anchored immutable identity anchors are preserved by the PURGE mechanism.

42. The network gateway device according to claim 41, further comprising an update protocol configured to include, in each firmware update check request, a monotonic counter value readfrom OTP-protected storage, and wherein an update server is configured to reject requests wherein the included monotonic counter value does not exceed the highest previously recorded value for the device, defeating replay of outdated firmware update offers.

43. The network gateway device according to claim 41, wherein firmware update check requests are transmitted to an update server on a schedule with a random time jitter derived from the hardware true random number generator, the random jitter being at least plus or minus two hours relative to a nominal check interval, to prevent synchronized update check requests from a fleet of devices and to avoid creation of a predictable outbound traffic pattern.

44. A method of session management for an administrative web interface of a network security device comprising: generating, using a hardware true random number generator, a session token of at least 128 bits entropy upon administrator authentication; binding the session token to the IP address of the authenticated administrator client; rotating the session token at intervals no greater than five minutes during an active session; and invalidating the session token upon detection of a request bearing the session token but originating from an IP address other than the bound IP address; wherein authentication of each request includes a constant-time comparison of the provided session token with the stored session token to prevent timing side-channel attacks.

45. The method according to claim 44, further comprising: generating a CSRF protection token as the HMAC of the session secret key over a concatenation of the session token and a nonce; binding the CSRF token to the specific session such that a CSRF token obtained from one session is invalid in any other session; and requiring a valid CSRF token for all administrative interface requests that modify device configuration.

46. A network gateway device comprising: a NAT session table comprising a total number of session entries; a first partition of the session table reserved for device-originated control-plane traffic comprising DNS resolution, NTP synchronization, syslog forwarding, firmware update, and administrative services, the first partition having a fixed maximum number of reserved entries; and a second partition of the session table available for traffic forwarded on behalf of the peripheral accessory, wherein the first partition is sized so that camera-side traffic volume cannot prevent device-originated control-plane services from creating NAT session entries.

47. A method of DNS proxying for an isolated network peripheral device comprising: intercepting, at a network gateway device, all DNS queries originating from a peripheral accessory on an isolated network port, regardless of the destination IP address specified in the DNS query; forwarding the DNS query to a configured upstream DNS resolver via a second network port; receiving a DNS response from the upstream resolver; removing from the DNS response allresource records whose type is other than A, AAAA, or CNAME; limiting the DNS response to a maximum number of answer records; and returning the filtered DNS response to the peripheral accessory; wherein the peripheral accessory is prevented from communicating directly with any DNS server.

48. A log-appending method for a network security device with no battery-backed RAM comprising: preparing a new log entry comprising a payload and an HMAC computed over a concatenation of a previous log entry HMAC and the current payload; writing the new log entry to a standby flash buffer; making the standby flash buffer active via an atomic pointer swap; and upon boot, verifying the HMAC chain from the beginning of the log and truncating the log at the first entry with an invalid HMAC, recording the truncation event as the next log entry.

49. The network security device according to claim 48, further comprising a write-once flash sector configured to record selected critical security events, including tamper detection events, repeated authentication failures, firmware update events, and secure boot verification results, wherein the write-once flash sector is not erased by a factory reset operation or a PURGE operation.

50. A network gateway device comprising a physical reset button and an LED indicator, wherein the reset button is configured to trigger different recovery operations based on button press duration including at least a first operation triggered by a first press duration, a second operation triggered by a second press duration longer than the first, and a third operation triggered by a third press duration longer than the second, and wherein the LED indicator changes its blink pattern at each press duration threshold while the reset button is held, providing visual confirmation of the selected recovery operation to an operator without requiring the operator to count elapsed seconds.

51. A method for randomizing heap placement in a resource-constrained real-time operating system comprising: at device boot, reading a random offset value from a hardware true random number generator; allocating a heap memory region larger than the minimum required heap size; selecting a starting address within the allocated region by adding the random offset to the base address of the allocated region, constrained to a boundary compatible with a hardware memory protection unit alignment requirement; and initializing the real-time operating system heap at the selected starting address; wherein the random offset provides at least four bits of entropy to defeat firmware exploitation techniques that rely on fixed heap base addresses.

52. The method according to claim 51, further comprising saving memory protection unit configuration to non-volatile registers or memory before entering a low-power processor mode and re-arming the memory protection unit configuration before any user task is permitted toexecute following wakeup from the low-power mode.

53. The network isolator according to claim 25, wherein the physical separation of the first input / output port from the second input / output port is implemented by one of: (a) independent network interface controllers connected to a common processor via independent bus paths with no shared frame-buffer memory between the first and second ports; (b) a single integrated circuit supporting both ports via independent MAC controllers with firmware enforcing that all traffic between the ports transits the controller's whitelist and filtering logic; (c) a single physical radio module supporting both ports as logically distinct virtual interfaces separated by distinct encryption domains; or (d) any combination of hardware and firmware measures that ensures no packet originating on the peripheral-accessory side can reach the computer network side without authorization by the controller, regardless of whether the first and second ports share a common circuit board, enclosure, or silicon substrate.