Communication system and method for improving resilience to jamming and radio frequency interference

Abstract

A communication system (100) comprises a plurality of drones (110), wherein each drone of the plurality of drones comprises a drone-borne base station and wherein, whilst air-borne, the drone-borne base stations are configured to communicate with at least one other drone-borne base station and function as a meshed network.

Description

COMMUNICATION SYSTEM AND METHOD FOR IMPROVING RESILIENCE TO JAMMING AND RADIO FREQUENCY INTERFERENCETechnical Field

[0001] The technical field relates generally to a communication system and method for improving the resilience of wireless communication networks to jamming and (hostile) radio frequency (RF) interference. In particular, the technical field relates to a system and method to provide alternative communications for so-called friendly forces operating in conflict zones, where Global Navigation Satellite System (GNSS) jamming and hostile RF interference are prevalent.Background

[0002] In recent years, third generation (3G) wireless communications have evolved to the longterm evolution (LTE™) cellular communication standard, sometimes referred to as 4thgeneration (4G) wireless communications. Both 3G and 4G technologies are compliant with third generation partnership project (3 GPP™) standards. 4G networks and phones were designed to support mobile internet and higher speeds for activities, such as video streaming and gaming.The 3 GPP™ standards are now developing a fifth generation (5G) of mobile wireless communications, which is set to initiate a step change in the delivery of better communications, for example powering businesses, improving communications within homes and spearheading advances such as driverless cars.

[0003] One of the potential technologies targeted to enable future cellular network deployment scenarios and applications is the support for wireless backhaul and relay links that enable flexible and very dense deployment of 5G-new radio (NR) cells without a need for densifying the transport network proportionately.

[0004] Due to the expected larger bandwidth available for NR, compared to long term evolved (LTE™) (e.g., mmWave spectrum) technologies, together with the native deployment of massive multiple-in / multiple-out (MIMO) or multi-beam systems in NR, opportunities have arisen todevelop and deploy integrated access and backhaul (IAB) links. It is envisaged that this may allow easier deployment of a dense network of self-backhauled NR cells in a more integrated manner, for example by building upon many of the control and data channels / procedures defined for providing access to UEs.

[0005] It is known that fixed cellular mesh-networks exist that can extend the range of 5G and implement IAB parts of the 5G specification by leveraging multiple nodes to distribute and route signals, making it ideal for rural and hard-to-reach areas. This allows for multiple individual nodes to be connected to each other simultaneously and being dynamically allocated to one or more other nodes at any specific time in order to grow or contract the network as ‘business need’ dictates.

[0006] The inventors have recognised and appreciated the potential benefits of extending (at least) 5G and 6G communications (and beyond) into previously under-utilised environments and / or difficult environments where there are unreliable communications, such as conflict zones, where it is known that various interference mechanisms exist. In conflict zones in particular, maintaining reliable and secure communication is critical for military and defence operations. Therefore, unwanted interference, which includes GNSS jamming and adversarial RF disruptions, can significantly degrade traditional communication networks. Thus, a system, devices and methods are needed for improving the resilience of wireless communication networks to jamming and (hostile) RF interference.Summary

[0007] In a first aspect, a communication system is described that comprises a plurality of drones, wherein each drone of the plurality of drones comprises a drone-borne base station and wherein, whilst air-borne, the drone-borne base stations are configured to communicate with at least one other drone-borne base station and function as a meshed network.

[0008] In this manner, employing a plurality of drone-borne base station configured to function as a drone-borne meshed network improves a resilience of wireless communication networks to jamming and (hostile) RF interference. In some examples, the drone-borne meshed network isused to complement an existing communication system, and in other examples the drone-borne meshed network is employed to provide communications in environments, e.g., hostile environments where no current communication system exists (other than satellite-based communications that can be jammed).

[0009] In some optional examples, the plurality of drone-borne base stations may be configured to form a meshed radio access network (RAN), and at least one drone-borne base station may be configured to communicate with a land-based base station. In some optional examples, the at least one drone-borne base station may be configured to communicate with the land-based base station that forms an integrated access and backhaul, IAB, communication link. In some optional examples, the IAB communication link may be configured as a secure communication link between the drone-based meshed network and a land-based communication network.

[0010] In some optional examples, the at least one drone-borne base station may comprise: a centralized unit, CU, a radio unit, RU, operably coupled to the CU, a distributed unit, DU, operably coupled to the CU and RU, and a learning processor operably coupled to the CU and RU and DU, wherein the learning processor is configured to perform radio frequency, RF, spectrum analysis and determine therefrom at least one of: frequency selection, power control, to be employed in the meshed network. In some optional examples, the learning processor may be configured to perform radio frequency, RF, spectrum analysis, the RU being configured to monitor RF conditions and the learning processor being configured to control one or more of the following in the meshed network: optimize route selection for data transmission; frequency hopping of drone-deployed base stations; use an available frequency that is unused.

[0011] In some optional examples, the learning processor may be configured to determine an adjustment of at least one transmission setting of at least one drone-borne base station, and the RU may be configured to transmit an instruction of the at least one transmission setting to the at least one drone-borne base station. In some optional examples, the instruction of the at least one transmission setting may comprise at least one of: a transmit power setting, re-orientation of at least one antenna that is transmit to the at least one drone-borne base station. In this manner, the drone-based meshed network may be adaptable to be re-configured in response to the variable communications and drone locations of the dynamic drone-based meshed network.

[0012] In some optional examples, the CU is configured to modify a geographical coverage shape of the meshed network following an activation or deactivation of at least one drone-based base station. In some optional examples, the learning processor may be configured to determine at least one of: frequency selection, power control, to be employed in the meshed network to counteract a determined radio interference. In some optional examples, the CU may be configured to perform network control technology of data transfer over an user plane, and signaling control over a control plane. In some optional examples, the CU may be configured to determine from multiple communication paths an alternative selected communication path in the meshed network in response to a failure of one or more drone-borne base station. In some optional examples, the meshed communication network is a fifth generation, 5G, or sixth generation, 6G, communication network.

[0013] In a second aspect, a method for improving the resilience of wireless communication network is described. The method comprises: affixing a plurality of base stations to a respective plurality of drones; making air-borne each drone of the plurality of drones; and whilst air-borne, and configuring a first drone-borne base station to communicate with at least one other drone- borne base station and function as a meshed network.Brief Description of the Drawings

[0014] Further details, aspects and embodiments will be described, by way of example only, with reference to the drawings. In the drawings, similar reference numbers are used to identify like, or functionally similar, elements. Elements in the figures are illustrated for simplicity and clarity and have not necessarily been drawn to scale.

[0015] FIG. 1 illustrates a drone-based deployable communication architecture providing a meshed network according to some example embodiments.

[0016] FIG. 2 illustrates a simplified 5G architecture configured to support IAB with a land- based structure or a moving drone, according to some example embodiments.

[0017] FIG. 3 illustrates a simplified flowchart 300 of a 5G network deployment process is illustrated, in accordance with some example embodiments.

[0018] FIG. 4 illustrates a block diagram of an IAB base station (or node), adapted in accordance with some example embodiments.

[0019] FIG. 5 illustrates an example of a neural network that may be employed as an artificial intelligence-based learning processor architecture to analyse RF spectrum, for example identifying RF space availability / suitability within a contested RF environment, according to some examples of the present invention.

[0020] FIG. 6 illustrates an example of a neural network that may be employed as an AI-ML based RF spectrum analysis system, according to some examples of the present invention.

[0021] FIG. 7 illustrates one example representation of an accelerometer arrangement, according to some example embodiments.

[0022] FIG. 8 illustrates one example of a beam forming arrangement, configured to support wireless communications by IAB base station (or node) of FIG. 4, according to some example embodiments.

[0023] FIG. 9 illustrates a simplified flowchart of a method for improving the resilience of wireless communication network, in accordance with some example embodiments.

[0024] Skilled artisans will appreciate that elements in the figures are illustrated for simplicity and clarity and have not necessarily been drawn to scale. For example, the dimensions and / or relative positioning of some of the elements in the figures may be exaggerated relative to other elements to help to improve understanding of various embodiments herein described. Also, common but well-understood elements that are useful or necessary in a commercially feasible embodiment are often not depicted in order to facilitate a less obstructed view of these various example embodiments. It will be further appreciated that certain actions and / or steps may be described or depicted in a particular order of occurrence while those skilled in the art will understand that such specificity with respect to sequence is not actually required. It will also be understood that the terms and expressions used herein have their ordinary technical meaning as is accorded to such terms and expressions by persons skilled in the technical field as set forth above except where different specific meanings have otherwise been set forth herein.Detailed Description

[0025] In order to improve the resilience of wireless communication networks to jamming and (hostile) radio frequency (RF) interference, particularly in conflict situations, examples herein- described propose combining and integrating a suite of technologies, deployed (and / or delivered) by drones, that establish a robust meshed communication network (e.g., a 5G meshed network) that is capable of operating in conflict situations.

[0026] In some examples, a plurality of drone-borne base stations are deployed, which in some examples are arranged to deliver a ‘pop-up’ network, such as a 5G meshed network, in any location that the drones are able to move into, to either function as a meshed network or to deliver base stations that are capable of forming a meshed network. It is envisaged that, whilst being air-borne, the drone-borne base stations are configured to function as a meshed network, such as a 5G meshed network. In some examples, the location of the plurality of drones, whilst being air-borne, is controlled, say, by a drone controller, which is arranged to move each of the drones individually to optimize the communication capabilities of the air-borne meshed network. In some examples, it is envisaged that the drone controller may be used in a process of delivering a ‘pop-up’ 5G network. In this ‘delivery’ example, it is envisaged that the drone-borne base stations functioning as a 5G meshed network may be unloadable from the transporting drones. In some examples, the drone-borne 5G meshed network may be achieved by deploying all elements of a 5G network that are capable of being carried on drones.

[0027] In some examples, the drone-borne 5G meshed network may be formed of multiple mobile terminal distributed units (mt-DU) connected to radio units (RUs) on each node (i.e., drone) to form a meshed network with a single centralized unit (CU) hosted on a (typically centrally-located) drone. Thus, in some examples, GnodeB hardware may include the RU, which in some examples may be formed using: (i) an Antenna array, (ii) a Radio Frequency Front End (RFFE), (iii) transceiver circuits (including frequency generation circuitry), and (iv) associated power and hardware equipment sufficient to send signaling through the antenna array and RFFE.

[0028] In some examples, it is envisaged that communications and / or data may be relayed via multiple routes in parallel essentially breaking the standard tree structure of a 5G IAB RAN and building it into the drone-borne 5G meshed network structure.

[0029] In some examples, it is envisaged that each of the drones may include a core, CU, DU and RU software stack, as well as a radio access network intelligent controller (RIC). The RIC, via O-RAN interfaces, can dynamically configure the RAN to have certain behavioral characteristics. In other examples, it is envisaged that the 5G meshed network may alternatively be deployed / delivered as a balloon-based infrastructure.

[0030] In some examples, it is envisaged that artificial intelligence (Al)-driven or machine learning-driven neural network model, sometimes referred to as a ‘learning processor’ (and corresponding) RF technology may be employed in the centralized unit (CU) hosted on a (typically centrally-located) drone, for dynamic and intelligent spectrum selection and power control, which may include monitoring RF conditions and optimizing route selection for data transmission, thereby ensuring reliable connectivity even in contested RF environments. In some examples, it is envisaged that the dynamic and intelligent spectrum selection may encompass frequency hopping of the drone-deployed base stations. In some examples, it is envisaged that the output may be arranged to control the corresponding RF technology of 5G base stations to facilitate optimized 5G transmission settings, which may be utilized effectively, in some examples, to enable either hard to detect low power 5G networks or networks that can use frequency or power settings that are unused, but identified faster as being available by the AL driven or machine learning-driven neural network model.

[0031] In some examples, the drone deployed 5G meshed network may be configured to provide 5G network capability and corresponding 5G services to a set of end-consumers, in a location where such service does not currently exist. Alternatively, in some examples, the 5G services may be provided in addition to existing civilian or military communication services. In this manner, the proposed solution may be used in remote locations as a stand-alone network without a need to connect to the wider internet in order to form a ‘backhaul’.

[0032] In some examples, the drone-deployed 5G meshed network may be configured to provide data only, data and voice, and / or voice only services to UEs. In some examples, the deployed 5G meshed network may be arranged to provide configurable levels of service depending on endservices required by the user. In some examples, the drone-deployed 5G meshed network may be arranged to provide specific service characteristics dependent on the user needs, be it Machine-to-Machine (M2M), Internet of Things (IOT) or person-to-person communication. Hereafter, the term ‘user’ is intended to encompass a device, a machine or a human user.

[0033] In some examples, the drone deployed 5G meshed network may be configured to include GnodeB software, employed to provide appropriate functionality and operation of the GnodeB hardware. In some examples, the drone-deployed 5G meshed network may utilize part of the 5G specification known as Integrated Access Backhaul (IAB) implemented on the GnodeB. In some examples, the IAB functionality may be implemented by using Frequency Division Duplex (FDD), using two frequencies, or Time Division Duplex (TDD) using one frequency noting that the option of either TDD or FDD is not IAB specific.

[0034] Operational Support System (OSS) tooling is employed in a remote data centre or operational headquarters to operationally support and manage the deployed platform for provisioning and management of services to User Equipment (UE) Devices. This has been developed to manage this drone-deployed 5G meshed network and the service deployed across the network. It is envisaged that the OSS tooling may be configured to perform one or more of: provision SIM(s) on the network to provide UE service; manage a ‘live’ network and the services deployed, operational using the drone-deployed network; monitor the deployment to provide an end-to-end service overview for the users of the service; service usage data collection for analysis and / or subsequent billing system integration(s); and provision, connection and management of the IAB mesh network.

[0035] Thus, in some examples, the OSS tooling may be arranged to include setting up an initial multi-node (meshed) network and may be arranged to dynamically modify the ‘shape’ of the network as nodes (i.e., drones) are activated in (introduced into) the network or deactivated from the network. In some examples, de-activated may include dynamic ‘removal’ of the node unexpectedly from the network, (for example as if the node is destroyed by military action). In some examples, management of an IAB mesh network includes specifically modifying the ‘shape’ of the mesh network to respond to customer demand for service being activated or deactivated as per their business need at the time. For example, the ‘shape’ of the mesh network may be modified as the network detects an increase in use, or a negative impact to the network (such as jamming). Thus, the OSS tooling is able to react / alert to negative impacts to the performance of the network with the ‘shape’ of the mesh network modified to account for extrademand or changes needed to the network topology. The term 'shape' of the mesh network refers to the dynamic topology of interconnected nodes, which can be adjusted by activating or deactivating nodes, changing transmission parameters, or re-orienting antennas (see FIG. 7) to optimize signal coverage and counteract jamming.

[0036] In some examples, it is envisaged that construction of the radio unit RU of the base station may include a physical fixing element to fix the RU to drones to facilitate air-borne travel of the 5G meshed network. In some examples, it is envisaged that the fixing to the transporting entity may be permanently attached or attached but removable from a transporting drone.

[0037] In some examples, it is envisaged that integration of positioning information via an information display / availability (via screen or IT interface) may be applied and connected to the transported radio unit. In some examples, it is envisaged that a GNSS resiliency Position Navigation and Time (PNT) concept may be employed to allow the radio units to maintain resilient clock and time measurements, thus allowing the network to function whilst in a GNSS denied environment. A GNSS-denied environment refers to areas where GNSS signals are deliberately jammed or blocked, thereby preventing traditional positioning methods. Thus, in some examples, it is envisaged that the 5G meshed networks may require very accurate timing in order to function properly. In this regard, it is envisaged that three individual Chip Scale Atomic Clocks (CSAC) may be bonded on one or more drone-based 5G base stations. By use of bonding of multiple clock data sources in this manner, a determination as to whether jamming exists, or a specific 5G base station / node is an active target for jamming of the GNSS signal may be obtained. In this context, data fusion encompasses taking all time inputs, and forming an accurate time from those that are available, thereby precluding a need to receive a jammed / GNSS-denied clock reference. In this manner, each drone-based 5G base station may be able to both broadcast Position, or Time to User Equipment (UE’s such as phones and other devices), which enables Navigation, and is able to determine any future base station geographical location by relay of time, whilst also concurrently supporting network communications. In this manner, it is possible to deliver low-power, safe, secure, high performance, robust and reliable operational timing.

[0038] In some examples, it is envisaged that a combination of the aforementioned technologies, enables a 5G meshed network to be constructed and used by drones and / or transported to any geographical location by drones where the 5G meshed network is able tofunction when an existing GNSS is suffering interference (e.g., using the herein-described PNT approach). In some examples, the 5G meshed network is able to identify what spectrum is best to use, and with what RF power (e.g., using the herein-described Al approach).

[0039] In some examples, it is envisaged that a combination of the aforementioned technologies, enables the deployed 5G meshed network to be able to be continuously updated as GNSS or RF spectrum interference changes. In some examples, it is envisaged that updates may dynamically redeploy (either automatically or via manned intervention) the geographical configuration of the network and RF signature that it uses.

[0040] Referring now to FIG. 1 , FIG. 1 illustrates a drone-based deployable communication system 100 providing a meshed network, according to some example embodiments. The dronebased deployable communication system 100 illustrates a plurality of drones 110, where a number of the plurality of drones 110 have been adapted to include (or configured to transport) base stations, such as 5G base stations. In the illustrated system, the plurality of drones 110 are configured in a meshed network 140. In some examples, the meshed network 140 supports both inter-base station communication 130 within the mesh and may be arranged to communicate 150 with a backhaul-supported CU-based drone base station 120 drone, using, say, IAB (which in some examples is connected to a backhaul land-based communication link (not shown)). In this manner, the IAB links may reduce reliance on traditional fiber or satellite-based backhaul. In some examples, each base station may be arranged to operate with peer-to-peer routing, thereby enabling data relay and self-healing in case of a node failure.

[0041] In some examples, with the plurality of drones 110 being configured in a meshed network 140, it is envisaged that the meshed network 140 may be configured to provide uninterrupted service by rerouting traffic around the mesh in case of a particular node (base station) failure. In this manner, the meshed network 140 facilitates multiple communication paths from the wireless communication device to the land-based communication network via one or more other base stations and, say, the CU-based drone base station 120. This arrangement allows for the selection of the most suitable communication path in case of one or more points of node failure.

[0042] In some examples, it is envisaged that the a relationship of communication between the plurality of drones 110 in the meshed network 140 may not be 1: 1. Thus, it is envisaged that the communication between the plurality of drones 110 may be 1:N, where N is to be defined dependent on the circumstance (for example N may be >1 and hence a meshed network 140 is formed). Thus, the flow of data between and management of the meshed network 140 may follow a 1: 1: 1 (e.g. a chain) data flow, so the CU-based drone base station 120 is able to see throughout the meshed network 140 via this chain, where all plurality of drones 110 are configured to disperse instructions. However, in some examples, it is envisaged that communications may be dispersed throughout the meshed network 140 in a 1:N:N nodes manner, for example depending upon mesh configuration at that point.

[0043] In some examples, the drone-based base station deployment may be arranged such that the meshed network 140 operates in two primary modes: (i) an in-flight mode, whereby the drone continuously provides 5G coverage while airborne; and (ii) a stationary deployment mode, whereby the drone lands and detaches a deployable base station on sea or land 160. Thus, the drones 110 may serve as a temporary communication node during transportation, and later be arranged to unload its payload (of base stations) as a stationary communication hub. In this manner, with respect to deploying communication capability in a conflict zone, it is envisaged that the provision of alternative communications is of great use to friendly forces to maintain communications despite various and diverse interference mechanisms.

[0044] In some drone-deployments, radio access network (RAN) technology and Integrated Access and Backhaul (IAB) may be used to provide a ‘wireless network over a specific geographical location’. IAB may be included to enable self-backhauling; however, in cases where backhaul is unnecessary, standalone operation without IAB may be supported. In some examples, a single land-based base station and a single relay drone-based (base station) may be employed to form the IAB link. In this manner, one drone-deployed base station may be configured as a donor IAB node (typically connected to a core network via satellite, fiber, or another medium), which is arranged to serve as an anchor drone-deployed base station. In this example, it is envisaged that one or more drone-deployed base station(s) may be configured as a relay IAB node, used to extend coverage by dynamically establishing new links and adjusting coverage based on network topology demands.

[0045] In this example, it is envisaged that UEs or land-based edge nodes may be arranged to interact with these drone-deployed base stations, requesting services, whilst benefiting from dynamic network reconfiguration. The 5G meshed network RAN may be deployed alongside network control technology (to allow for transfer of data over the 5GUser Plane, and control of signaling over the 5G Control Plane), where the control plan for the meshed network is managed from a drone such as a CU.

[0046] In this example, the wireless mesh network is arranged to support IAB network functionality. One of the main objectives of IAB is to provide radio access network (RAN)-based mechanisms to support dynamic route selection to accommodate short-term blocking and transmission of latency-sensitive traffic across communication links. This objective is also relevant to resource allocation (RA) between access and communication (e.g., backhaul) links under half-duplexing constraints. In the NR standard, there are three RA modes defined, namely time division multiplex (TDM), frequency division multiplex (FDM) and space division multiplex SDM (e.g., beam-based operation).

[0047] In some examples, a 5G meshed network 140 as described herein, may be deployed as an alternative to any existing communication network. In some examples, a 5G meshed network as described herein, may integrate a number (and in some instances all) of the following: (i) a 5G meshed Radio Access Network (RAN) technology, (ii) an integrated access and backhaul (IAB), (iii) a resilient Position, Navigation, and Timing (PNT) methodology, and (iv) Artificial Intelligence (Al)-driven RF spectrum analysis.

[0048] In some examples, the 5G base stations deployed within the drone-borne meshed network 140 interact dynamically to form a resilient and adaptive communication system. In some examples, these interactions may be governed by one or multiple factor(s), such as interference from other RF emitters, power or range needs, etc.

[0049] In some examples, the drone-based base stations may be deployed as semi-autonomous units, capable of: autonomous boot-up and self-initialization when powered on. The drone-based base stations may also be arranged to dynamically re-configure the mesh ‘on-the-fly’, thereby enabling drones or mobile units to be dynamically repositioned based on network demands. In some examples, the drone-based base stations are configured to support remote Over-the-Air(OTA) updates, thereby ensuring continuous improvement without requiring manual intervention.

[0050] These functionalities ensure that the 5G meshed network remains operational despite hostile or unpredictable conditions. By integrating these design principles, the 5G base stations within this meshed network ensures maximum resilience, adaptability, and operational reliability, even in the most challenging environments.

[0051] In a wireless meshed network 140, where communications are often passed between many different entities (sometimes herein referred to as ‘nodes’), data packets typically have an option of being transferred between a source entity or node and a destination entity or node via a number of alternative routes. Each respective communication between entities on the route is generally termed a ‘hop’. It is known that in wireline networks a typical routing metric that is used is the number of hops, i.e., routes are selected that have the minimum number of hops, as this is often the most time-efficient and cost-efficient route. This works well in wireline networks because in a wireline system either a link between two nodes is available, in which case losses are minimal, or the link is not available and no data can pass. Thus, since neighboring nodes are by definition ‘ 1 ’ hop away, the cost metric is always ‘1’.

[0052] However, using the ‘number of hops’ metric to select a preferred route in wireless networks can result in a poor performance, if routes that involve long paths with high loss rates are selected. Typically, wireless mesh networks use a re-transmission scheme at the medium access control (MAC) layer (e.g., 802.11 WiFi™) and the long links selected by the use of such a ‘minimum hop count’ metric will require a high rate of re-transmissions in order to send the data (i.e., they have a low delivery rate). Therefore, in many circumstances it can be found that it is better to select a route with more hops, where each of the hops requires fewer re-transmissions (i.e., each of the hops has a higher (successful) delivery rate). In the example, wireless mesh network, the route selection is more complicated based on the respective communication link efficiencies that are, in large part, dependent upon the speed and accuracy of the antenna beamforming in response to sensors.

[0053] In order to create a meshed network 140 it is necessary to first discover devices that can provide mesh connectivity. One known mechanism to achieve this is for all nodes within themesh to take a turn transmitting beacon information, as is currently the case in for example 802.11 IBSS (independent BSS also known as ‘ad hoc mode of operation’). A device can then find mesh connectivity by simply performing a passive scan attempting to find beacons of an appropriate variety. Thus, in some examples, it is envisaged that multiple IAB base stations (or nodes) functioning as IAB nodes, a communication system and methods relating to mesh usage and communications within an environment or between an environment and a land-based communication network have been described, wherein the aforementioned disadvantages with prior art arrangements have been substantially alleviated.

[0054] Referring now to FIG. 2, a simplified land-based 5G architecture 200 is configured to support an Integrated Access and Backhaul (IAB) network, linking the centralized unit (CU)- based drone base station 120 of FIG. 1 with IAB capability with various land-based wireless communications, is illustrated according to some examples. In this manner, the IAB links may reduce reliance on satellite-based backhaul.

[0055] Here, a first 5G land-based base station 222 is operably connected to the drone base station with AIB capability (e.g., CU-based drone base station 120 in FIG. 1), preferably with secure communications. The first 5G land-based base station 222 supports communications within a coverage area 224, including communication support for a plurality of mobile (or fixed) wireless communication units, sometimes referred to as a terminal device, such as a user equipment UE 226. In 5G, the UE 226 is able to support traditional Human Type Communications (HTC) or the new emerging Machine Type Communications (MTC). The simplified 5G architecture 200 includes a second 5G base station 212 supporting communications within a coverage area 214, including communication support for a plurality of mobile (or fixed) wireless communication units, such as UE 216. A wireless backhaul connection 233, generally an Xn (based on X2) interface connects the first 5G base station 222 with the second 5G base station 212. The first 5G base station 222 is also connected to the core network via a more traditional wired connection, such as fibre 234.

[0056] In this regard, in an IAB scenario, the drone base station with AIB capability (e.g., CU- based drone base station 120) is considered a donor IAB node and node A (i.e., first 5G basestation 222) together with node C (i.e., second 5G base station 212) are identified as relay IAB nodes. Thus, in this manner, a drone-deployed meshed network (e.g., meshed network 140 in FIG. 1) is able to support communications between communication devices in a hostile location (for example that did not have any recognizable communication) and a civilian (or military) communication system 200.

[0057] Referring now to FIG. 3, a simplified flowchart 300 of a 5G network deployment process is illustrated, in accordance with some example embodiments. The simplified flowchart 300 starts at 302 with a first radio unit being activated. The simplified flowchart 300 proceeds at 304 with the activated first deployed radio unit performing an assessment of a status of a Global Navigation Satellite System (GNSS) status, and in particular whether the GNSS is compromised. In this regard, the first deployed radio unit checks a CSAC predictive time to a GPS time in order to provide an indication of GPS spoofing. The same is true if the CSAC derived position is vastly different from a corresponding determined GPS position, where the first deployed radio unit flags an error. If, at 304, the first deployed radio unit determines that the GNSS is compromised, the first deployed radio unit uses PNT at 306 and proceeds to 308. Alternatively, if, at 304, the first deployed radio unit determines that the GNSS is not flowchart proceeds direct to 308. In this manner, the first deployed radio unit a node or more with a sat position fix, we can then relay around GPS jammers and provide the PNT signal using the local CSACs.

[0058] At 308, the Al model (for example as described with reference to FIG. 6) selects a frequency and a power for the first deployed radio unit to use. Therefore, if the RF sniffer detects that a jammer (or the user) is active, then the Al model works out clear (non-jammed) frequency bands that should be used and advantageously, in real-time, changes the frequency band and / or transmit power used by the first deployed radio unit. This approach is more applicable for higher-power chips, in order to provide a macro-cell performance (albeit only for a short time, say due to thermal effects). In this manner, an override possibility for jamming events may be achieved for short periods of time.

[0059] The simplified flowchart 300 proceeds at 310, by establishing a local 5G communication bubble in order to provide locally provisioned services. In this example, a local 5G communication bubble is a network that includes a core, CU, DU and RU in a single instance,which may or may not include an Internet communication link. The simplified flowchart 300 proceeds at 312, by an Operator at a Headquarters or base location transmitting a location and frequency to a next node, for example a start of a relay of both PNT data as well as IAB data. Here, a default bootstrapped two-node network would be created, and via AI / ML as described herein, specific characteristics based on the outcome of this Al result would then be assessed as to the configuration required to be given to the network would be given.

[0060] The simplified flowchart 300 proceeds at 314, by deploying additional 5G base station (nodes) sequentially (with N instances of 5G base station. Once the assumed RF two-node network is established, an RF coverage map based on these network characteristics is known; and within the RF coverage, a third node may be deployed from the HQ or base location, joining the network whenever at a suitable position.

[0061] The simplified flowchart 300 proceeds at 316, by monitoring and adjusting the network topology. For example, during deployment, various conditions may change that mean the network needs to change. This may include: (i) the unforeseen drop of a connected node; (ii) Network node performance; (iii) a change in GNSS interference levels or resumption of GNSS service; (iv) interference and jamming of specific nodes or frequencies. In this manner, the monitoring tooling of the IAB network will alert the Al control or HQ and base staff that the network needs to be modified. In some examples, it is envisaged that this can be achieved via automated or manual processes.

[0062] Referring now to FIG. 4, a block diagram of a drone-based base station 400 (which in some examples encompasses a base-station (removably) affixed to a drone) is illustrated, in accordance with some example embodiments. In some examples, the drone-based base station 400 is constructed of a modular design to enable rapid deployment and adaptability. In some examples, the modular design may be divided into physical entities (or integrated circuits or devices) that offer scalability, where components, such as RF front ends, antennas, power modules, etc. may be upgraded or swapped based on operational needs. In some examples, the modular design offers interchangeability, whereby standardized interfaces allow different hardware and software components to be interchanged based on mission requirements (e.g.,adding new spectrum bands). In some examples, the modular design offers a compact and lightweight design, particularly useful for aerial and mobile platforms for drones, minimizing weight whilst maintaining power efficiency. In some examples, as illustrated, the modular design may include a radio unit (RU) 450, sometimes referred to as ‘radio head’, which is configured to handle the physical layer of communications in the protocol stack and typically arranged to house the RF front end, antennas, beamforming circuits, transceiver devices and other circuits for signal transmission and reception.

[0063] In some examples, as illustrated, the modular design may include a distributed unit (DU) 460, which may handle signal processing and lower-layer protocol functions of the protocol stack, such as Radio Link Control (RLC) and Medium Access Control (MAC) layer. In some examples, as illustrated, the modular design (and notably for the CU-based drone base station 120 in FIG. 1) may include a centralized unit (CU) 470, which is arranged to manage higher- layer functions of the protocol stack such as Service Data Adaption Protocol (SDAP), Packet Data Convergence Protocol (PDCP) and Radio Resource Control (RRC), including network coordination (for example to communicate with the IAB network 200 in FIG. 2), handover management, and interference mitigation. In some examples, as illustrated, the modular design may include a power system, which may include battery and / or solar power backup, optimized for extended operation. In some examples, as illustrated, the modular design may be configured to support future systems and technologies, such as allowing integration of 6G and beyond without significant hardware overhauls.

[0064] In some examples, the control (e.g., in the controller 414 or CU 470 may be managed by machine language (ML) or artificial intelligence (Al) of a neural network, such as described with reference to FIG. 6. In some examples, the ML / Al of the neural network may be configured to control functionality of the drone, for example in terms of changing frequency, location, transmit power, antenna beam forming, etc. on the drone or via secure communication to / from a centrally-located drone, for example to meet specific mission aims. In this context, it is envisaged that the meshed network may include laying down a signal only over a specific geographical area, or having high power beams for backhaul over the front line and then low power beams to users at the front line to minimalize the RF footprint.

[0065] In some examples, the IAB base station (or node) 400 may be simplified to function as a repeater, without the suite of control functions that are contained in a fully-functional IAB base station. The IAB base station (or node) 400 contains an antenna array 402, for receiving transmissions, coupled to a beam-forming circuit 404, which controls the antenna elements contained in the antenna array 402, with regard to orientation, power, etc. The beam-forming circuit 404 is coupled to an antenna switch 403 that provides isolation between receive and transmit chains within the IAB base station 400.

[0066] In the RU 450, one or more receiver chains, as known in the art, include receiver frontend circuitry 406 (effectively providing reception, filtering and intermediate or base-band frequency conversion). The receiver front-end circuitry 406 is coupled to a signal processing module 408 (generally realized by a digital signal processor (DSP) and illustrated as forming at least a part of a DU 460). A skilled artisan will appreciate that the level of integration of receiver circuits or processors or components may be, in some instances, implementation-dependent.

[0067] The controller 414 maintains overall operational control of the drone-based base station 400. The controller 414 is also coupled to the receiver front-end circuitry 406 and the signal processing module 408. In some examples, the controller 414 is also coupled to a frequency generation circuit 417 and a memory device 416 that selectively stores operating regimes, such as decoding / encoding functions, synchronization patterns, code sequences, and the like. A timer 418 is operably coupled to the controller 414 to control the timing of operations (e.g., transmission or reception of time-dependent signals) within the drone-based base station 400.

[0068] As regards the transmit chain, this essentially includes an input circuit 420 configured to receive information from one or more sensors 490, connected to signal processor 408. The signal processor 408 is coupled in series through transmitter / modulation circuitry 422 and a power amplifier 424 to the antenna array 402, which in some examples is configured as a steerable beam-forming antenna. The transmitter / modulation circuitry 422 and the power amplifier 424 are operationally responsive to the controller 414. The signal processor 408 in the transmit chain may be implemented as distinct from the signal processor in the receive chain. Alternatively, a single processor may be used to implement a processing of both transmit and receive signals, as shown in FIG. 4. Clearly, the various components within the drone-based base station 400 can berealized in discrete or integrated component form, with an ultimate structure therefore being an application-specific or design selection.

[0069] In some examples, it is envisaged that drone-based base station 400 may be dynamically reconfigured, based on signal strength variations. Thus, in some examples, the signal processor 408 includes an active signal strength tracking circuit 430, to receive, process and analyze received signals. In this manner, the active signal strength tracking circuit 430 of the drone-based base station 400 is able to influence the operational circuits of the drone-based base station 400 to improve (or provide) reliable and robust communication by constantly monitoring and optimizing the signal strength and signal quality of received signals.

[0070] In some examples, the signal processor 408 includes a Feed-forward Kinematic Model 432. The Feed-forward Kinematic Model 432 is operably coupled to the input circuit 420 to receive sensor information. In this manner, sensors placed on the drone-based base station 400, for example when it is functional on a drone, are utilized to determine, say, a vertical inclination and axial rotation of the antenna array 402. In some examples, this information is fed into the feed-forward kinematic model 432, which allows for proactive adjustment of the antenna's position in advance of use, thereby enabling timely and accurate communication.

[0071] In some examples, the signal processor 408 includes a predictive model circuit 434, which is configured to consider the environmental state of the antenna structure, to anticipate changes in the antenna's aspect (for example when it is functional on a drone) and relative to user equipment that the drone-based base station 400 is (or wants to) communicating with. The predictive capability provided by the predictive model circuit 434, coupled to antenna beamform controller 436 ensures that adjustments are made in a timely manner, thereby aligning the antenna for optimal communication. In a first example, it is envisaged that the predictive model circuit 434 may be implemented using, say, a number of phase locked oscillators linked to the motions of the drone involved. In some examples, the predictive model circuit 434 is configured to determine how the effects of the prevailing conditions of the drone-based base station 400, i.e., changes identified in the sensor information, will affect the change in angle and direction of the antenna array 402. In some examples, this information may include modelling the inertial response times, which can factor in windage.

[0072] In some examples, the signal processor 408 includes a drone orientation / positioning circuit 435, which is configured to receive sensor information via the input circuit 420. In this manner, the sensor information enables the drone orientation / positioning circuit 435 to determine a current orientation of the drone-based base station 400, which may include an inertial measurement based on, say, information from three axis accelerometers, and / or three axis gyroscope, and / or three axis magnetometer, as described in FIG. 7. In some examples, the sensors 490 may be configured to identify or determine changes in position (e.g. a global navigational satellite system (GNSS) or a global positioning service (GPS) location of the drone); and identify or determine effects of wind on the structure (e.g., by employing a wind speed sensor). In some examples, the sensors and / or accelerometers are configured to assess movement of the drone-based base station 400 (and by implication movement of the antenna beam former, where the movement includes angle acceleration and wind speed that can cause side loading.

[0073] In accordance with some example embodiments, the signal processor 408 and transceiver (e.g., transmitter / modulation circuitry 422) of the drone-based base station 400 may be configured to communicate with another drone-based base station within the meshed network, or a land-based base station / gNB, say via the IAB system 200 of FIG. 2.

[0074] In accordance with some example embodiments, the drone-based base station 400 incorporates a plurality of sensors to determine a real-time orientation and azimuth of the dronebased base station 400, and in particular the antenna array 402. A signal processor 408 is connected to the plurality of sensors and the antenna array 402, in a form of an antenna beamforming system. The signal processor 408 receives the sensor information and utilizes it to adapt the antenna array 402, for example the configuration of the beam-forming system for optimal communication with either the land-based communication network or other drone-based base stations.

[0075] In some examples, a power control mechanism 440 may be arranged to control the output power of the drone-based base station 400. In some examples, the power control mechanism 440 may employ intelligent power management to provide energy-efficient operation, which is important for drones with limited battery capacity. In some examples, the IAB base station (ornode) 400 may support alternative power sources, such as solar panels or fuel cells for extended field use.

[0076] In some examples, the drone-based base station 400 includes a neural network functioning as an artificial intelligence (Al)-machine language (ML) 445 (in a form of a learning processor) may be implemented on-board the deployed drone base station. Here, the Al 445 may be arranged to run on edge computing hardware within the drone or base station. In this example, the Al 445 may enable real-time decision-making without relying on external networks. In other examples, it is envisaged that the Al 445 (e.g., in a form of a learning processor) may be implemented external to the deployed drone base station, for example cloud-based Al 445 processing. In these examples, cloud-based Al training may be employed, which may periodically include new models being trained using aggregated data from multiple missions. In some examples, updated Al models may be pushed to deployed drones or base station, for example via over the air (OTA) updates.

[0077] FIG. 5 illustrates an example of a neural network 500 that may be employed as an artificial intelligence / machine learning (AI-ML)-based learning processor architecture for (inter aha) dynamic and intelligent spectrum selection and power control, say within a contested RF environment, according to some examples described herein. The neural network 500 continuously learns from real-time signal conditions, historical interference patterns, and regulatory spectrum constraints to ensure optimal network performance. In some examples, the neural network 500 is structured as a deep learning model, typically a convolutional neural network (CNN) or recurrent neural network (RNN), trained to dynamically and intelligently identify optimal spectrum and power configurations, for example within a contested RF environment, in order to broadcast suitable 5G signals. In some examples, it is envisaged that the output may include, or facilitate, optimized 5G transmission settings, which may be utilized effectively in some examples to enable either hard to detect low power 5G networks or networks that can use frequency or power settings that are unused, but identified faster as being available by the neural network Artificial Intelligence.

[0078] In some examples Al-driven spectrum selection may include frequency hopping, which is arranged to dynamically shift a selected frequency to a least-congested and least jammed frequency band or ensure a low probability of interception and detection (LPI / LPD). In some examples Al-driven spectrum selection may include adaptive bandwidth allocation, which may include dynamically assigning a higher bandwidth to critical nodes and reduce bandwidth for low-priority services. In some examples Al-driven spectrum selection may include AI-Based Beamforming Optimization, which may be configured to adjust a beam direction and power to attempt to obtain an optimal signal-to-noise ratio (SNR).

[0079] In some examples, the Al may be arranged to support a power control mechanism, which may be configured to provide an energy-efficient transmission, for example to minimize power consumption while maintaining strong signals. In some examples, the Al may be arranged to ensure interference avoidance, for example to reduce out-of-band emissions that may cause selfinterference in a dense 5G network. In some examples, the Al may be arranged to support a stealth mode of operation, whereby, if operating in a hostile environment, the neural network 500 may set the transmission power to just above the noise floor, thereby making detection difficult

[0080] In some examples, as understood by a skilled artisan, the neural network 500 may be continuously trained and updated, for example using reinforcement learning. In some examples, it is envisaged that Al training data sources that may be used for training the neural network 500 may include one or more of the following: RF spectrum measurements from operational deployments, simulated adversarial jamming scenarios, field-collected interference reports from conflict zones. In some examples, the neural network 500 may employ a reward system, whereby higher scores are allocated for reducing jamming impact, maximizing bandwidth, and optimizing power usage. Similarly, it is envisaged that the neural network 500 may employ a penalty system, whereby low scores are allocated for unnecessary power usage or inefficient spectrum allocation.

[0081] In some examples described herein any reference to Al encompasses a mathematical methodology to extract information from signals and / or images that may be similar to that used in cognitive functions in humans associated with learning or pattern recognition, in a supervised or unsupervised deep learning setting. Here, flexible adaption is able to extract informationpreviously not necessarily recognized or hypothesized or based on manually derived algorithms specific to the measurement of defined parameters. In the former case, Al in the setting of unsupervised learning is able to exploit serendipity in much the same way as scientific discoveries are made by chance.

[0082] In some examples described herein, any reference to ML, as distinct from Al, may be used as a mathematical methodology that refers to mathematical and statistical methods, which can without supervision, analyse and identify a specific outcome related to, in the case of examples described herein, a performance of analysing an identification of radio frequency (RF) space availability / suitability, say within a contested RF environment. In some examples, it may apply to algorithms and iterative methodologies in order to arrive at an optimal identification of a set task. In some examples, it may be used for data mining. For simplicity of terminology, however, any reference to Al examples of the invention described herein encompass supervised and unsupervised learning, as well as Machine Language.

[0083] In some examples, an artificial neural network is a network of artificial neurons and nodes meant for solving learning processor-type problems, such as (Al) problems. The connections of the neuron are weighted to model excitatory or inhibitory connections and summed. An activation function defines the amplitude of the output. It is envisaged in some applications of the invention that such artificial neural networks may be used for predictive modelling and adaptive control. In some examples, these neural networks may also be trained via a dataset, such as the examples described herein. Self-learning or unsupervised learning, resulting from analysis of datasets, can occur within the network, which allows the network to derive conclusions from a seemingly unrelated information set.

[0084] In some examples, the dataset that the Al algorithms will use as a learning platform will include, but is not limited to, data obtained from multiple data sources to assess the real-time and historical RF conditions, for example within a contested RF environment and a non-contested environment. The variables derived from the techniques discussed herein can be derived using conventional analysis or derived using Al methodologies. In some examples, these inputs may include one or more of: signal interference data, spectrum availability, historical usage patterns.

[0085] Signal interference data may be used for jamming detection, e.g., identifying adversarial RF jamming attempts and natural signal interference using measured Signal-to-Interference-plus- Noise Ratio (SINR) and Bit Error Rate (BER). Signal interference data may be used to identify interference patterns, for example by analyzing frequency spectrum congestion by detecting unwanted signals from hostile sources, nearby 5G networks, or legacy systems. Signal interference data may also be used to identify multipath effects, for example by detecting reflection, diffraction, and scattering effects that may degrade signal quality.

[0086] In some examples, spectrum availability as an input may be used to control dynamic frequency selection, for example using scans available from potential spectrum bands (sub-6 GHz, mmWave) in real time. In some examples, spectrum availability as an input may be used to ensure regulatory compliance with frequency licensing. In some examples, spectrum availability as an input may be used to improve bandwidth utilization, for example by identifying underutilized or congested channels, as well as tracking current bandwidth demand from connected users.

[0087] In some examples, historical usage patterns may be used as an input to enable the Al model to learn from past data, for example by analyzing past spectrum assignments and their effectiveness and / or accounting for time-of-day patterns and mission-specific RF conditions. In some examples, historical usage patterns may be used for predictive analysis, whereby Al predicts upcoming spectrum congestion based on past events. In this instance, it is envisaged that the historical usage patterns may be include seasonal interference modeling (e.g., expected jamming in military zones).

[0088] Referring now to FIG. 5, an example of a neural network 500 that may be employed as a learning processor, such as an artificial intelligence (Al)-based architecture to analyse the timing of RF signals is illustrated according to some examples of the present invention. In some examples, the example neural network 500 may comprise a convolutional neural network 500, which applies a series of node mappings 580 to an input 510, which ultimately resolves into an output 530 consisting of one or more values, from which at least one of the values is used by theneural network 500, for example an Al-based architecture. The example convolutional neural network 500 comprises a consecutive sequence of network layers (e.g. layers 540), each of which consists of a series of channels 550.

[0089] In some examples, the neural network 500 architecture includes an input layer that is configured to accept multi-dimensional sensor data (e.g., signal strength, interference levels, bandwidth demand, historical trends as described herein). Furthermore, it is envisaged that the neural network 500 may also include hidden layers. It is envisaged, in some examples, that the hidden layers may include feature extraction, which may be configured to identify signal anomalies, jamming attempts, and usable frequency bands and / or identify pattern recognition, which may use historical data to predict optimal spectrum allocations. It is envisaged, in some examples, that the hidden layers may include power optimization, which may determine a minimum necessary transmission power to maintain coverage whilst avoiding detection (if required) and / or include a decision layer, which may output real-time recommendations for spectrum usage and power levels.

[0090] The channels 550 are further divided into input elements 560. In this example, each input element 560 may store a single value. Some (or all) input elements 560 in an earlier layer are connected to the elements in a later layer by node mappings 580, each with an associated weight. The collection of weights in the node mappings 580, together, form the neural network model parameters 547. For each node mapping 580, the elements in the earlier layer are referred to as input elements 560 and the elements in the output layer are referred to as the output elements 570. An element may be an input element to more than one node mapping, but an element is only ever the output of one node mapping function 520.

[0091] In order to calculate the output 530 of the convolutional neural network 500 the system first considers the input layer as the earlier layer. The layer(s) to which the earlier layer is connected by a node mapping function 520 are considered in turn as the later layer. The value for each element in later layers is calculated using the node mapping function 520 in equation [1], where the values in the input elements 560 are multiplied by their associated weight in the node mapping function 520 and summed together.Node mapping function 520: d = A(wadx a + wbdX b + wcdX c) [1]

[0092] The result of the summing operation is transformed by an activation function, ‘A’ and stored in the output element 570. The convolutional neural network 500 now treats the previously considered later layer(s) as the earlier layer, and the layers to which they are connected as the later layers. In this manner the convolutional neural network 500 proceeds from the input layer 540 until the value(s) in the output 530 have been computed.

[0093] In some examples, it is envisaged that Al-driven or Machine Learning-driven Neural Network Model may be employed for dynamic and intelligent spectrum selection and power control, for example within a contested RF environment, in order to control corresponding RF technology for each deployed node (5G base station) when broadcasting suitable 5G signals. In some examples, it is envisaged that the output 530 may include, or facilitate, optimized 5G transmission settings, which may be utilized effectively in some examples to enable either hard to detect low power 5G networks or networks that can use frequency or power settings that are unused, but identified faster as being available by the neural network Artificial Intelligence.

[0094] In some examples, it is envisaged that the control of corresponding RF technology for each deployed node (5G base station) may include dynamic frequency assignment, which in some examples may include assigning specific spectrum blocks to each base station. In this manner, the dynamic frequency assignment may be used to avoid congested frequencies and adapt the spectrum frequency selection in a real-time manner in response to jamming threats.

[0095] In some examples, it is envisaged that the control of corresponding RF technology for each deployed node (5G base station) may include power level adjustments, which may include adjusting one or more deployed node’s (5G base station’s) power output based on, say, environmental conditions. In this manner, the power level adjustments may ensure energy efficiency and stealth operations when necessary.

[0096] In some examples, it is envisaged that the control of corresponding RF technology for each deployed node (5G base station) may include adaptive beamforming and antenna configuration, as described with respect to FIG. 5. In some examples, the control of corresponding RF technology for each deployed node (5G base station) may include determiningoptimal antenna orientation based on drone movement and user demand. In some examples, the control of corresponding RF technology for each deployed node (5G base station) may include dynamically steering beams to maintain stable communication links.

[0097] In some examples, it is envisaged that the control of corresponding RF technology for each deployed node (5G base station) may include network topology adjustments. For example, if interference is detected, it is envisaged that the Al-driven or Machine Learning-driven Neural Network Model output 530 may be employed to reconfigure the mesh network, for example the mesh network in FIG. 1. In this manner, the Al-driven or Machine Learning-driven Neural Network Model output 530 may be employed to activate new nodes (5G base stations) in order to reroute traffic or shift existing nodes (5G base stations) to multi-hop relay communication if direct communication links are compromised.

[0098] In examples of the invention, the convolutional neural network 500 may be trained. In some examples of the invention, the training of the convolutional neural network 500 may entail repeatedly presenting training data, for example from a sniffed RF model or training data for PNT and in order to calibrate the 5G base station’s clocks, as the input 510 of the convolutional neural network 500, in order to analyse the performance of the 5G system. In some examples of the invention, an optimisation algorithm may be used to reduce a loss function, for example by measuring how much each node mapping 580 weight contributed to the loss, and using this to modify the node mapping functions 520 in such a way as to reduce the loss. Each such modification is referred to as an iteration. After a sufficient number of iterations, the convolutional neural network 500 can be used to analyse the function from an input of timing of RF signals data.

[0099] In some examples of the invention, the large number of model parameters 547 used in the convolutional neural network may require the device to include a memory 590. The memory 590 may be used to store the training data 515, the model parameters 547, and any intermediate results 593 of the node mappings.

[0100] Thus, in the proposed 5G system, input data (a training dataset, clinical dataset, model parameters or intermediate results) is fed to the learning processor neural network in aformat that fits the input matrix. Nodes are mapped in a specific way that is adapted to the purpose of the system (forming e.g. a convolutional neural network). The information is gradually reduced through a series of interconnected input / output elements to generate the final output.

[0101] This Al -powered spectrum optimization model ensures that the drone-borne 5G network can adapt to any RF environment, avoid interference, and maintain resilient connectivity even under hostile conditions.

[0102] Referring now to FIG. 6, an example of a neural network that may be employed as an Al -ML based RF spectrum analysis system 600 is illustrated, according to some examples of the present invention. In some examples, the AI-ML based RF spectrum analysis system 600 may employ deep learning models, such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), or Transformer-based models, in order to process RF signals. The neural network implemented as an AI-ML based RF spectrum analysis system 600 is designed to analyze the availability and suitability of RF space in environments where multiple signals, interference, and adversarial factors may impact reliable communications. In some examples, the neural network implemented as an AI-ML based RF spectrum analysis system 600 may be configured to operate as a cognitive radio system that dynamically assesses RF spectrum conditions using machine learning techniques.

[0103] The AI-ML based RF spectrum analysis system 600 starts at 610 with RF data input to the AI-ML neural network / data acquisition and key characteristics from RF signals are extracted at 620. At 610, the AI-ML based RF spectrum analysis system 600 first gathers raw RF data using sensors and spectrum analyzers. The RF data input features may include: frequency, bandwidth, and power levels, where the AI-ML system 600 is configured to detect open RF bands and occupied frequency bands. In some examples, the RF data input features at 610 may include interference patterns, which may be used to identify jamming, interference, or spectrum congestion. In some examples, the RF data input features at 610 may include modulation & signal characteristics, which may be used to differentiate between friendly, neutral, andpotentially adversarial signals. In some examples, the RF data input features at 610 may include time and geolocation data, which may use factors in temporal RF usage patterns and geospatial spectrum characteristics.

[0104] At 620, feature extraction is performed by the Al -ML neural network. At 630, the Al -ML neural network performs neural network processing to analyze features. The process then diverges into three distinct analytical pathways.

[0105] A first path includes, at 640, a spectrogram analysis being performed, which may use time-frequency representations to detect signal anomalies. The system analyzes RF spectrograms, identifying different frequency components and interference patterns. Thereafter, at 670, a real-time RF spectrum map is output, which may highlight available and occupied RF bands and understanding of frequency utilization.

[0106] A second path includes, at 650, the Al -ML system 600 being configured to perform Al -ML processing and pattern recognition for threat identification, configured to detect anomalies or recurring signal behaviors. Thereafter, at 680, threat identification and mitigation is performed, for example using a probability of spectrum availability based on historical data and / or pattern recognition results to classify potential threats, which may identify adversarial jamming, unauthorized transmissions, and suggests countermeasures.

[0107] A third path includes, at 660, the Al -ML system 600 being configured to perform reinforcement learning (RL) for adaptive spectrum allocation, whereby the system learns optimal spectrum usage strategies over time. Thereafter, at 690, adaptive spectrum allocation may include dynamically allocating frequency resources (e.g., optimal RF bands for communication) to avoid interference and optimize usage.

[0108] In some examples, it is envisaged that the RF spectrum map at 670 provides foundational data to both threat identification at 680 and adaptive spectrum allocation at 690. The RF spectrum map at 670 may provide real-time spectrum usage insights, helping Threat Identification at 680 detect unauthorized transmissions or anomalies. The RF spectrum map at 670 also aids in adaptive spectrum allocation at 690 by identifying which frequency bands are available or congested. This allows for dynamic reallocation of spectrum resources based onreal-time RF conditions. If a threat (e.g., jamming, unauthorized access, interference) is detected at 680, the system may need to modify spectrum allocation to avoid compromised frequencies. Thus, the threat identification at 680 provides this intelligence to adaptive spectrum allocation at 690, thereby ensuring spectrum is assigned securely and efficiently in order to mitigate or remove interference.

[0109] In other examples (not shown) the Al -ML system 600 may be configured to perform anomaly detection and signal classification, which may identify unknown RF signals that might be interfering. Here, one final output may be to provide spectrum suitability scoring, which may rate each RF band’s usability based on interference, availability, and reliability.

[0110] In some examples, it is envisaged that additional enhancements may be performed by the Al -ML system 600 in a contested RF environment (e.g., military or emergency communications). Here, the AI-ML system 600 may enhance survivability by detecting and avoiding jamming signals, for example where the AI-ML system 600 may rapidly shift frequencies to mitigate disruptions. Furthermore, the AI-ML system 600 may use cognitive adaptive learning in order to continuously update its RF models in response to new signals. Additionally, the AI-ML system 600 may use multi-agent collaboration in order to use AI-ML agents in multiple locations in order to share real-time RF spectrum intelligence.

[0111] In some examples, the AI-ML neural network (sometimes termed a ‘learning processor’) provides an advanced RF spectrum analysis tool that is configured, in accordance with the examples described herein, capable of one or more of the following: identifying RF space availability dynamically, predicting interference & jamming attempts, providing adaptive spectrum usage recommendations, and ensuring reliable communication in contested environments.

[0112] Referring now to FIG. 7, one example representation of an accelerometer 700 arrangement located on a drone-deployed base station (e.g., drone-deployed base station 110 in FIG. 1 or base station 400 in FIG. 4) is illustrated, according to some example embodiments Accelerometers are being incorporated into an ever-expanding variety of electronic systems. Forexample, MEMS (MicroElectroMechanical System) accelerometers now are commonly used in automotive systems and in consumer and industrial electronics.

[0113] According to this example embodiment, accelerometer 700 includes a three-axis accelerometer, with a first accelerometer (e.g., accelerometer 712, FIG. 7) configured to sense the magnitude of a force along the x-axis 702, a second accelerometer (e.g., accelerometer 713, FIG. 7) configured to sense the magnitude of a force along the y-axis 704, and a third accelerometer (e.g., accelerometer 714, FIG. 7) configured to sense the magnitude of a force along the z-axis 706. Alternatively, accelerometer 700 may include a one-axis or a two-axis accelerometer, or no accelerometer, per se, at all whereby a different sensor arrangement may be used. In addition or alternatively, accelerometer 700 may include a three-axis gyroscope in an embodiment, which includes a first gyroscope (e.g., gyroscope 722, FIG. 7) configured to sense the rate of rotation around the x-axis 702, a second gyroscope (e.g., gyroscope 723, FIG. 7) configured to sense the rate of rotation around the y-axis 704, and a third gyroscope (e.g., gyroscope 724, FIG. 7) configured to sense the rate of rotation around the z-axis706. Alternatively, accelerometer 700 may include a one-axis or a two-axis gyroscope, or not use a gyroscope, per se, at all. In some example embodiments a gyroscope is used with a Kalman filter (sometimes referred to as an accelerometer). In this context, the gyroscope measures a rate of change of angular velocity, as compared to an absolute angular velocity. In an alternative example embodiment, sensors may be used to detect an angle using a pendulum and accelerometer.

[0114] In addition or alternatively, accelerometer 700 may include a three-axis magnetometer in an embodiment, which includes a first magnetometer (e.g., magnetometer 732, FIG. 7) configured to sense the intensity of a magnetic field along the x-axis 702, a second magnetometer (e.g., magnetometer 733, FIG. 7) configured to sense the intensity of a magnetic field along the y-axis 704, and a third magnetometer (e.g., magnetometer 734, FIG. 7) configured to sense the intensity of a magnetic field along the z-axis 706. Alternatively, accelerometer 700 may include a one-axis or a two-axis magnetometer, or no magnetometer, per se, at all. In some examples, it is envisaged that accelerometer 700 may include one or more additional or different types of sensors, as well (e.g., pressure sensors, temperature sensors, chemical sensors, and so on). As used herein, the term “sense,” when referring to an action performed by a transducer,means that the transducer produces an electrical signal indicating the magnitude of an acceleration, a force or a field impinging upon the transducer, or the rate of an angular acceleration (rotation) experienced by the transducer.

[0115] FIG. 7 is a simplified block diagram of an accelerometer device 700 that includes multiple transducers 712-714, 722-724, 732-734, in accordance with one example embodiment. More specifically, device 700 includes a three-axis accelerometer 710, a three-axis gyroscope 720, and a three-axis magnetometer 730, in an embodiment. Accelerometer 710 includes x-axis accelerometer 712 configured to sense the magnitude of acceleration (e.g., due to gravity) along an x-axis of the device 700 (e.g., x-axis 702, FIG. 1), y-axis accelerometer 713 configured to sense the magnitude of acceleration (e.g., due to gravity) along a y-axis of the device 700 (e.g., y-axis 704, FIG. 1), and z-axis accelerometer 714 configured to sense the magnitude of acceleration (e.g., due to gravity) along a z-axis of the device 700 (e.g., z-axis 706, FIG. 1). Accelerometers 710 may function to measure static acceleration (e.g., for quantifying the tilt angles experienced by device 700) and / or dynamic accelerations (e.g., for quantifying dynamic acceleration of device 700 in a particular direction). According to alternate embodiments, device 700 may include only a subset of accelerometers 712-714, or may have no accelerometer at all. Gyroscope 720 includes x-axis gyroscope 722 configured to sense the rate of rotation around the x-axis of the device 700, y-axis gyroscope 723 configured to sense the rate of rotation around the y-axis of device 700, and z-axis gyroscope 724 configured to sense the rate of rotation around the z-axis of device 700. According to alternate embodiments, accelerometer 700 may include only a subset of gyroscopes 722-724, or may have no gyroscope at all. Magnetometer 730 includes x-axis magnetometer 732 configured to sense the intensity of a magnetic field along the x-axis of the device 700, y-axis magnetometer 733 configured to sense the intensity of a magnetic field along the y-axis of the device, and z-axis magnetometer 734 configured to sense the intensity of a magnetic field along the z-axis of the device 700. According to alternate embodiments, device 700 may include only a subset of magnetometers 732-734, or may have no magnetometer at all. Accelerometer 700 may include one or more additional or different types of sensors, as well (e.g., pressure sensors, temperature sensors, chemical sensors, and so on).

[0116] According to one example embodiment, accelerometer 700 may further include multiplexers 716, 726, 736, 750 (MUXs), gain and filter circuitry 760, analog-to-digital converter 770 (ADC), and microcontroller 780 (or another suitable control and / or processing component). In some examples, microcontroller 780 may include memory 782 (e.g., data and instruction registers, Flash memory, read only memory (ROM), and / or random access memory (RAM)), although all or portions of memory 782 may be external to microcontroller 780, as well. Memory 782 may be used to store a variety of types of persistent and temporary information or data, such as executable software instructions (e.g., instructions associated with implementation of the various embodiments), data (e.g., data generated based on transducer output signals (“transducer data” herein), configuration coefficients, and so on), device parameters, and so on. Further, microcontroller 780 may include an external interface 784, which is configured to facilitate communication between microcontroller 780 and an external electrical system (not illustrated in FIG. 7) through conductive contacts 792, 793, 794, 795 exposed at an exterior of accelerometer 700. For example, external interface 784 may include a Serial Peripheral Interface (SPI), an Inter-Integrated Circuit (I2C), another type of interface, or a combination of various types of interfaces. Accelerometer 700 may further include conductive contacts (not shown) configured to convey voltage references (e.g., power and ground) to various nodes and / or internal electrical components.

[0117] Under control of microcontroller 780, MUXs 716, 726, 736, 750 enable one signal at a time to be selected from accelerometers 712-714, gyroscopes 722-724, and magnetometers 732-734, and provided to gain and filter circuitry 760. Gain and filter circuitry 760 may be configured to apply, say, a pre-determined gain to the selected signal (e.g., as specified by a gain value or code selected during a calibration process), and to filter the signal in order to eliminate spurious, out-of-band signal components, in order to filter out signal components caused by undesired mechanical vibrations, and / or to filter out electrical noise (e.g., from other circuitry of accelerometer 700). The resulting analog signal is provided to ADC 770, which samples the analog signal and converts the samples into a sequence of digital values. Each digital value indicates the magnitude of an acceleration, a force or a magnetic field impinging upon the accelerometer 700 (e.g., when the value corresponds to a signal from theaccelerometer 710 or gyroscope 720, respectively), or the rate of acceleration around an axis (e.g., when the value corresponds to a signal from the magnetometer 730).

[0118] According to an example embodiment, accelerometers 712-714 may be MEMS accelerometers, for example. Typically, a MEMS accelerometer includes a mass with a first electrode, which is suspended over a second electrode that is fixed with respect to the accelerometer substrate. The first electrode, the second electrode, and the air gap between the first and second electrodes form a capacitor, and the value of the capacitor is dependent upon the distance between the first and second electrodes (i.e., the width of the air gap). As the mass is acted upon by an external force (e.g., gravitational force), the first electrode may move closer to or farther from the second electrode, and thus the capacitance value may decrease or increase, accordingly. It is envisaged that other accelerometer configurations also are possible, including configurations in which signals indicating multiple capacitances are generated. According to an example embodiment, each accelerometer 712-714 produces an electrical signal indicating the value of such capacitance(s).

[0119] According to an example embodiment, device 700 may include MUX 716, as mentioned previously, and may further include capacitance-to-voltage (C-to-V) converter 718. The electrical signals indicating the capacitance values associated with accelerometers 712- 714 are provided to MUX 716, which may selectively provide one of the signals to C-to-V converter 710 based on a control signal from microcontroller 780. C-to-V converter 718 is configured to convert the capacitance value into a voltage signal on line 719, where the voltage signal has a magnitude that is proportional to the capacitance value associated with the accelerometer 712-714 selected by MUX 716. In an alternate embodiment, device 700 may include a C-to-V converter for each accelerometer 712-714, and a MUX instead could be used to select a voltage signal output from one of the C-to-V converters.

[0120] According to an example embodiment, gyroscopes 722-724 may be MEMS gyroscopes, for example. Typically, a MEMS gyroscope includes a moving mass with an electrode (“movable electrode”), which is suspended adjacent to a second electrode (“fixed electrode”) that is fixed with respect to the gyroscope substrate. When the substrate is rotated about an axis perpendicular to the direction of motion of the moving mass, the mass experiencesa force in the third orthogonal direction of magnitude 2mv l, where m is the mass, v the velocity, and fl is the rotation rate.

[0121] Referring now to FIG. 8, one example of a beam forming circuit 403, configured to support wireless communications by drone-based base station 400 of FIG. 4, according to some example embodiments. In this manner, the drone-based base station 400 of FIG. 4 is able to adapt to its movement within the drone-based meshed network, as well as its own orientation / tilting and positioning to assist with communications to / from other communication units (e.g., other drone-based base stations).

[0122] In some examples, an inverse fast fourier transform (in a receive sense) function or fast fourier transform (in a transmit sense) function transforms each received input orthogonal frequency division multiplex (OFDM) symbol into a time signal. By nature of the IFFT function, the number of time samples is exactly the same as the number of subcarriers. The next operation is converting a (complex) digital signal into a (complex=IQ) analog signal in a DAC (also not shown) and upconverting to a double-sideband RF signal (no longer complex) in an up-mixer. The up-mixer uses a Local Oscillator (LO) signal as a time reference. When an n-channel analog beam former is being used, the upconverted RF signal 810 is split into n copies using RF splitter 812. At least the phase (using phase adjustment circuit 830, 832, 834, 836) and potentially the amplitude (using amplitude adjustment circuit 840, 842, 844, 846) of each signal can be adjusted, typically on a symbol-by-symbol basis or on a ‘multiple of symbols’ basis. This allows the beam direction to be adapted / updated. Before supplying the signals Y1 ... Yn to the antennas 850, 852, 854, 856, a power amplifier may be inserted, such as power amplifier 424 in FIG. 4 (not illustrated in this figure for clarity purposes only).

[0123] In some examples, depending upon the required total range of beam directions to be generated by beam forming circuit 403, those skilled in the art will appreciate that antennas 850, 852, 854, 856 can take a variety of forms. In some examples simple vertical dipoles or equivalent horizontally-polarised elements may be suitable, where the direction range is smaller examples could include patch antennas or dipoles with reflecting elements behind.

[0124] Thus, in some examples, the antenna elements may be implemented using any suitable techniques, e.g., a two-element yagi, sheet / mesh reflectors, corner reflectors - as with patches, where a designer is able to choose a radiation pattern to suit. In some examples, it is also envisaged that horizontally polarized antennas and omni coverage in azimuth may be used, which are much less common, e.g., using crossed dipoles or cloverleafs.

[0125] The implementation of the beam control is achieved using a beam-index 870 controlled by antenna beam-form controller 436. For each new beam direction, a reasonable number of antenna elements are typically controlled with many control bits, even if the number of beam directions is limited.

[0126] At a receive side, the signal processing is in reversed sequence of the aforementioned transmit operation. The received antenna signals are first amplified in a low noise amplifier (LNA) (again not illustrated in this figure for clarity purposes only). Next, a phase (using phase adjustment circuit 830, 832, 834, 836) and potentially the amplitude (using amplitude adjustment circuit 840, 842, 844, 846) can be modified to control the maximum sensitivity direction of the received beam. After summing up the signals in an RF combiner (a reverse operation of the RF splitter 812), the RF signal is down-converted to complex IQ baseband signals before being sampled in an ADC. After the ADC, a FFT is performed, the frequency representation of the received symbols is retrieved and in some instances the signal is then equalized. Since the gain and phase of each subcarrier signal typically suffers various offsets from losses and various reflections between transmit and receive antenna, correction by an equalizer may be required before QPSK or QAM symbols can be determined successfully.

[0127] Although some example embodiments have been described with respect to an analog beam-forming system, it is envisaged that other controllable beam-forming systems can be used, e.g., a digital beam-former.

[0128] Referring now to FIG. 9, a simplified flowchart 900 of a method for improving the resilience of wireless communication network is illustrated, according to some examples described herein. The flowchart 900 includes, at 910 affixing a plurality of base stations to a respective plurality of drones. At 920, the flowchart 900 includes making air-borne each drone of the plurality of drones. At 930, flowchart 900 includes, whilst air-borne, configuring a firstdrone-borne base station to communicate with at least one other drone-borne base station and function as a meshed network.

[0129] It will be appreciated that, for clarity purposes, the above description has described example embodiments with reference to different functional units and processors. However, it will be apparent that any suitable distribution of functionality between different functional units or processors, for example with respect to the signal processor may be used without detracting from the example embodiments described. For example, functionality illustrated to be performed by separate processors or controllers may be performed by the same processor or controller. Hence, references to specific functional units are only to be seen as references to suitable means for providing the described functionality, rather than indicative of a strict logical or physical structure or organization.

[0130] Aspects may be implemented in any suitable form including hardware, software, firmware or any combination of these. The example embodiments may optionally be implemented, at least partly, as computer software running on one or more data processors and / or digital signal processors or configurable module components such as field programmable gate array (FPGA) devices. Thus, the elements and components of an embodiment may be physically, functionally and logically implemented in any suitable way. Indeed, the functionality may be implemented in a single unit, in a plurality of units or as part of other functional units.

[0131] Although examples have been described in connection with some embodiments, it is not intended to be limited to the specific form set forth herein. Rather, the scope of the present invention is limited only by the accompanying claims. Additionally, although a feature may appear to be described in connection with particular embodiments, one skilled in the art would recognize that various features of the described embodiments may be combined. In the claims, the term ‘comprising’ does not exclude the presence of other elements or steps.

[0132] Furthermore, although individually listed, a plurality of means, elements or method steps may be implemented by, for example, a single unit or processor. Additionally, although individual features may be included in different claims, these may possibly be advantageously combined, and the inclusion in different claims does not imply that acombination of features is not feasible and / or advantageous. Also, the inclusion of a feature in one category of claims does not imply a limitation to this category, but rather indicates that the feature is equally applicable to other claim categories, as appropriate.

Claims

Claims:

1. A communication system (100) comprises: a plurality of drones (110), wherein each drone of the plurality of drones (110) comprises a drone-borne base station and wherein, whilst air-borne, the drone-borne base stations are configured to communicate with at least one other drone-borne base station from the plurality of drones (110) and function as a meshed network.

2. The communication system (100) of Claim 1 wherein the plurality of drone-borne base stations is configured to form a meshed radio access network (140), RAN and at least one drone- borne base station (120) is configured to communicate with a land-based base station.

3. The communication system (100) of Claim 2 wherein the at least one drone-borne base station (120) is configured to communicate with the land-based base station and form an integrated access and backhaul, IAB, communication link.

4. The communication system (100) of Claim 3 wherein the IAB communication link is configured as a secure communication link between the drone-based meshed network and a land- based communication network.

5. The communication system (100) of any preceding Claim wherein at least one drone- borne base station (120) comprises: a centralized unit, CU, a radio unit, RU, operably coupled to the CU, a distributed unit, DU, operably coupled to the CU and RU, and a learning processor operably coupled to the CU and RU and DU, wherein the learning processor is configured to perform radio frequency, RF, spectrum analysis and determine therefrom at least one of: frequency selection, power control, to be employed in the meshed network.

6. The communication system (100) of Claim 5 wherein the learning processor being configured to perform radio frequency, RF, spectrum analysis comprises the RU beingconfigured to monitor RF conditions and the learning processor being configured to control one or more of the following in the meshed network: optimize route selection for data transmission; frequency hopping of drone-deployed base stations; use an available frequency that is unused.

7. The communication system (100) of Claim 5 or Claim 6 wherein the learning processor is configured to determine an adjustment of at least one transmission setting of at least one drone- borne base station, and the RU is configured to transmit an instruction of the at least one transmission setting to the at least one drone-borne base station.

8. The communication system (100) of Claim 7 wherein the instruction of the at least one transmission setting comprises at least one of: a transmit power setting, re-orientation of at least one antenna that is transmit to the at least one drone-borne base station.

9. The communication system (100) of any of preceding Claims 5 to 8 wherein the CU is configured to modify a geographical coverage shape of the meshed network following an activation or deactivation of at least one drone-based base station.

10. The communication system (100) of any of preceding Claims 5 to 9 wherein the learning processor is configured to determine at least one of: frequency selection, power control, to be employed in the meshed network to counteract a determined radio interference.

11. The communication system (100) of any of preceding Claims 5 to 10 wherein the CU is configured to perform network control technology of data transfer over an user plane, and signaling control over a control plane.

12. The communication system (100) of any of preceding Claims 5 to 11 wherein the CU is configured to determine from multiple communication paths an alternative selected communication path in the meshed network in response to a failure of one or more drone-borne base station. re13. The communication system (100) of any preceding Claim wherein the meshed communication network is deployed as an alternative to an existing communication network or in addition to the existing communication network.

14. The communication system (100) of any preceding Claim wherein the meshed communication network is a fifth generation, 5G, or sixth generation, 6G, communication network.

15. A method (900) for improving the resilience of wireless communication network comprising: affixing (910) a plurality of base stations to a respective plurality of drones; making air-borne (920) each drone of the plurality of drones; and whilst air-borne, configuring (930) a first drone-borne base station to communicate with at least one other drone-borne base station and function as a meshed network.

Images