System for automated, middleware-based intelligent data mapping between heterogeneous ERP systems and cloud platforms

Abstract

A system for automated, middleware-based intelligent data mapping between heterogeneous enterprise resource planning (ERP) systems and cloud platforms, the system comprising the following: an ERP data acquisition unit with a variety of physical communication ports, selected from Ethernet and serial ports, a network interface controller physically connected to the communication ports, and a DMA (Direct Memory Access) controller configured to transfer raw ERP data packets into a volatile buffer memory without processor intervention; a Semantic Mapping Unit consisting of a Field Programmable Gate Array (FPGA) configured as a lexical analyzer to analyze ERP schema definitions and extract entity descriptors, and a Tensor Processing Unit (TPU) configured to evaluate the semantic similarity between heterogeneous schema elements based on ontological models stored in a non-volatile memory repository; a transformation logic unit comprising a parallelized vector processor configured to perform matrix-based transformations, field normalizations and structural reorganization of ERP data into a harmonized format, and a rule application circuit configured to perform compliance-based anonymization, unit conversions and value scaling prior to cloud distribution; a cloud interface unit comprising a protocol adapter bank with encoders, translators, and message queue writers, wherein the cloud interface unit is coupled with a cryptographic controller configured with a hardware public key infrastructure (PKI) circuit to generate digital signatures and verify integrity codes for all outgoing payloads; and a Feedback Learning Control Unit implemented in the firmware, wherein the engine is electrically coupled to the Semantic Mapping Unit and the Transformation Logic Unit and is configured to monitor error logs, success rates and acknowledgment codes received from the cloud platforms and to update the semantic and transformation rules stored in the non-volatile memory repository accordingly.

Description

Field of invention

[0001] The present invention relates generally to integration technologies for enterprise resource planning (ERP) and, in particular, to a system for automated intelligent data mapping that enables interoperability between heterogeneous ERP systems and cloud computing platforms such as Oracle. The invention lies at the interface between middleware hardware and software-driven mapping engines. Data schemas, transaction logs, and resource descriptors from different ERP systems are captured, normalized, transformed, and exported into standardized formats that can be used by cloud-based enterprise applications. Background of the invention

[0002] Enterprise resource planning (ERP) systems are used across industries to manage business processes, including finance, supply chain, human resources, manufacturing, and customer relations. However, companies often operate multiple ERP systems from different vendors, with varying architectures and schema models. These ERP systems may use incompatible data structures, communication protocols, and transaction logic, making seamless data interoperability difficult.

[0003] Current integration practices rely on manual schema mapping, proprietary connectors, or third-party integration services. These methods are often time-consuming, error-prone, and expensive, and they do not dynamically adapt to evolving business rules or ERP version updates. Furthermore, as organizations increasingly migrate workloads to cloud platforms, there is a pressing need for automated middleware systems that can harmonize heterogeneous ERP data flows, perform intelligent mapping using semantic insights, and ensure compatibility with cloud-native storage and processing layers.

[0004] Existing solutions are predominantly software-defined and only partially anchored at the device level, leading to bottlenecks in performance, latency, and security. Therefore, a technical gap exists for a dedicated, hardware-anchored system that implements intelligent mapping capabilities in conjunction with middleware controllers, thus providing high-speed transfer, schema harmonization, and adaptive transformation into cloud-compatible user data.

[0005] In the ever-evolving landscape of enterprise technology, integrating heterogeneous enterprise resource planning (ERP) systems into cloud platforms has become one of the biggest technical challenges for businesses worldwide. ERP systems are large-scale, integrated applications for managing core business functions such as finance, supply chain management, human resources, manufacturing, and customer relationship management. Over the years, various vendors, including SAP, Oracle, Microsoft Dynamics, Infor, and several open-source platforms, have developed ERP systems with unique data models, database schemas, communication protocols, and transaction logics. As companies expand through mergers, acquisitions, and partnerships, they often run multiple ERP systems concurrently, leading to significant heterogeneity in the organization's IT environment.While the rise of cloud computing offers scalable storage, enhanced analytics, and flexible processing capabilities, bridging ERP systems with cloud platforms requires intelligent mechanisms for data harmonization and transformation. Without such mechanisms, organizations face significant inefficiencies, inconsistent reporting, and operational bottlenecks.

[0006] Existing solutions for ERP cloud integration can essentially be divided into three categories: manual schema mapping, proprietary connectors, and Integration Platforms as a Service (iPaaS). With manual schema mapping, specialists analyze the ERP source schemas, identify corresponding fields in the target system, and write transformation scripts or queries to reconcile mismatched fields. This approach is very labor-intensive and error-prone, as the mapping often involves hundreds or thousands of fields across various ERP modules. Even the slightest change in the source or target schema can invalidate previous mappings and necessitate repeated work. Furthermore, manual mapping is not scalable and therefore unsuitable for companies with frequently evolving business models or multiple ERP implementations in different regions.

[0007] Proprietary connectors offered by ERP and middleware vendors attempt to automate integration by providing pre-built templates or connectors for common ERP-to-cloud scenarios. For example, connectors can enable direct integration between a specific ERP vendor's application and a particular cloud storage or analytics service. While these connectors reduce initial implementation time, their vendor-specific nature limits their versatility. Organizations using multiple ERP systems from different vendors often find that proprietary connectors cannot provide a unified solution. Furthermore, vendor-provided connectors frequently become obsolete when either the ERP or cloud platform undergoes a version upgrade, leading to maintenance issues.Another disadvantage is that connectors typically only perform syntactic transformations and not semantic interpretation. This means that fields, despite their structural mapping, may not correctly represent their meanings, leading to inconsistent data interpretation on the cloud side.

[0008] Integration Platforms as a Service (iPaaS) are gaining popularity as a cloud-based approach to ERP integration. These platforms offer a visual or low-code environment where developers can configure data pipelines, apply transformation rules, and manage data transfer workflows. They support multiple connectors, provide monitoring dashboards, and sometimes include built-in transformation capabilities. Despite these advantages, iPaaS platforms face significant limitations when it comes to handling ERP-specific complexity. The semantic richness of ERP data, including multi-level hierarchies, nested objects, and context-dependent meanings, is difficult to capture with generic transformation rules. Therefore, iPaaS often requires extensive customization, negating the promised simplicity.Furthermore, reliance on cloud-hosted integration platforms can raise security and compliance concerns, particularly for organizations handling sensitive financial or healthcare data, as the data must pass through third-party integration servers before reaching the target cloud. Latency also becomes an issue when large ERP datasets are transferred across remote iPaaS nodes, resulting in delays in near real-time analytics and reporting.

[0009] Another existing approach is the use of ETL (Extract-Transform-Load) pipelines. Traditional ETL tools extract data from ERP databases, apply transformation scripts to adapt it to the target schema, and load the processed data into cloud warehouses. While effective for batch workloads, ETL pipelines struggle with the real-time transactional data that modern businesses increasingly rely on. For example, a company might need real-time synchronization between its on-premises ERP system and a cloud-based analytics dashboard. ETL's batch-oriented approach leads to delays and, consequently, outdated insights. Furthermore, ETL processes often lack adaptive intelligence; once a mapping rule is defined, the pipeline cannot automatically adapt to schemas. This rigidity necessitates frequent manual intervention and undermines automation.

[0010] Some organizations attempt to address this issue by employing Enterprise Service Bus (ESB) architectures. These architectures connect ERP systems to a central bus via standardized messaging protocols such as SOAP or REST, which then forwards messages to the cloud. While ESBs offer a more systematic integration framework, they do not automatically resolve the semantic mapping problem. An ESB can reliably deliver messages, but if these messages are not harmonized at the semantic level, cloud platforms may receive inconsistent or incomplete data. Furthermore, ESBs increase the architectural complexity and introduce an additional infrastructure layer that requires maintenance, patching, and scaling. Performance bottlenecks can occur when multiple high-throughput ERP sources compete for the limited bus capacity.

[0011] Another class of solutions focuses on application programming interfaces (APIs). Many modern ERP systems offer APIs for data access, while cloud platforms provide APIs for data ingestion. Companies often create custom middleware layers that query ERP APIs, perform transformations, and transfer data to cloud APIs. While this offers flexibility, it results in high development and maintenance costs. Each API integration requires careful programming, testing, and monitoring. In heterogeneous environments with multiple ERP vendors, this approach is poorly scalable, as each ERP-cloud pair requires a tailored integration effort. Furthermore, API limits imposed by ERP vendors present practical constraints, especially when synchronizing large volumes of data.

[0012] Previous research has attempted to use semantic web technologies and ontologies for ERP integration. Theoretically, modeling ERP data elements as ontology classes and assigning them to a standardized ontology allows for semantic alignment. However, the practical application of such ontology-based approaches is limited by the complexity of maintaining large ontologies and the computational effort of the reasoning engines. These systems often operate at the software level, leading to performance degradation and limited acceptance in real-world enterprise environments where latency and throughput are critical.

[0013] A recurring drawback of existing solutions is the lack of hardware acceleration. Most current integration technologies are software-defined and run on standard servers or cloud instances. While this architecture is flexible, it cannot achieve the rapid, low-latency transformation required by organizations processing terabytes of ERP data daily. Furthermore, purely software-based solutions lack the necessary hardware-level security controls to ensure data integrity and compliance with regulations such as GDPR, HIPAA, or SOX. Given increasing regulatory scrutiny, organizations cannot afford middleware that does not provide auditable, secure, and verifiable integration paths.

[0014] Another major problem with existing solutions is their lack of adaptability. ERP systems and cloud platforms are constantly evolving, frequently releasing version upgrades, introducing new modules, or discarding old ones. A static mapping, once defined, quickly becomes obsolete. Traditional integration solutions require consulting teams to review mappings, update pipelines, and revalidate workflows. This reimplementation cycle is costly and time-consuming. Globally operating companies cannot afford such repeated disruptions, as they impact not only their IT systems but also their business continuity. The lack of self-learning and adaptive intelligence in existing middleware solutions makes them unsuitable for modern, dynamic business environments.

[0015] Scalability also remains a bottleneck. While small businesses can manage with simple connectors or manual mappings, multinational corporations operating across multiple continents struggle with significantly larger data volumes. Financial transactions, supply chain updates, and customer orders can generate millions of data records daily. Existing solutions suffer under high utilization, either experiencing performance degradation or requiring costly scaling of cloud-based integration services. For companies with business-critical operations, such as those in banking, healthcare, or aviation, the lack of a guaranteed scalable, real-time, and accurate ERP cloud integration poses unacceptable risks.

[0016] In addition to the technical disadvantages, there are also costs involved. Proprietary connectors and iPaaS subscriptions often come with ongoing licensing fees that increase with data volume, leading to rising costs. Custom API-based integrations require internal development teams or external consultants, which increases operating costs. ETL and ESB solutions require continuous monitoring, infrastructure investments, and maintenance contracts. Therefore, the total cost of ownership for existing ERP integration strategies often exceeds initial projections, especially as companies expand or undertake digital transformation initiatives.

[0017] Despite decades of development in ERP integration, no existing solution adequately meets the combined requirements for automation, semantic intelligence, hardware acceleration, security, adaptability, and scalability. Companies must compromise, opting for either rigid but reliable ETL pipelines, limited proprietary connectors, or flexible but costly iPaaS systems. This persistent gap underscores the need for a novel system that embeds intelligent mapping capabilities in a dedicated, hardware-based middleware device. Such a device would not only accelerate transformation processes but also provide adaptive semantic mapping, robust security, and protocol flexibility, thus overcoming the limitations of current approaches. Objectives of the invention

[0018] The invention aims to provide a system and device architecture that can autonomously map, transform, and forward ERP data to cloud platforms without human intervention. A further aim is to provide a hardware-based middleware device with specialized mapping engines, semantic processors, and cloud protocol adapters that improve the speed and reliability of enterprise integration. In addition, the invention aims to minimize errors caused by manual schema interpretation and introduce machine-driven mapping that continuously learns from transaction history and regulatory constraints. Summary of the invention

[0019] The invention describes a system for automated, middleware-based, intelligent data mapping between heterogeneous ERP systems and cloud platforms. The system comprises a middleware hardware device (MHD) installed in enterprise data centers. The MHD includes specialized units such as an ERP data acquisition unit, a semantic mapping unit, a transformation logic unit, and a cloud interface unit.

[0020] The ERP data acquisition unit captures structured and semi-structured raw data packets from multiple ERP sources using physical connectors, dedicated communication ports, and bus-level intercept controllers. The semantic mapping unit, implemented with reconfigurable FPGA-based lexical analyzers, extracts schema elements, transaction descriptors, and entity relationships from the incoming ERP data streams. The transformation logic unit applies hardware-accelerated matrix transformations, normalization rules, and AI-based ontological matching to convert ERP-specific data structures into a cloud-compatible data model. Finally, the cloud interface unit encapsulates the transformed payload into API-compatible or message queue-compatible packets for transmission to cloud storage and processing services.

[0021] The system also features a persistent knowledge repository in secure, non-volatile storage. This repository stores schema mappings, bug fixes, and regulatory metadata to optimize future transformations. A hardware-based cryptographic controller ensures that all data exchanges are authenticated and their integrity verified before being sent to the cloud.

[0022] The main objective of the present invention is to provide a system and a device that enable seamless, automated, and intelligent data mapping between heterogeneous ERP systems and cloud platforms, thereby eliminating the inefficiencies and risks associated with manual schema matching and proprietary connectors. The invention aims to create a hardware-based middleware structure that not only ingests and interprets diverse ERP data streams but also transforms them into harmonized, cloud-compatible user data with minimal human intervention. A further objective is the integration of semantic intelligence into the mapping process, enabling the system to recognize, align, and match meaningfully equivalent but structurally different data elements from various ERP vendors to ensure consistency and accuracy in the cloud environment.

[0023] Another objective of the invention is to provide a middleware device with integrated hardware acceleration for the rapid transformation, normalization, and semantic analysis of ERP data. By incorporating FPGA-based lexical analyzers, tensor processing units, and cryptographic controllers, the invention aims to overcome the performance limitations of purely software-based solutions. This hardware-based approach ensures low latency, high throughput, and the reliable execution of complex mapping operations, which are essential for companies processing large volumes of data in real time. Furthermore, the invention aims to integrate advanced security at the hardware level to ensure encrypted transmission, tamper-proof integrity checks, and compliance with regulatory requirements during ERP cloud integration.

[0024] Another objective of the invention is to enable adaptability and continuous learning in the mapping process. ERP systems and cloud platforms evolve rapidly, with frequent version updates and schema changes, and static integration methods often become obsolete. The invention addresses this challenge by embedding a feedback-driven learning engine that dynamically refines mapping rules based on real-time performance metrics, error logs, and confirmation signals from cloud systems. This ensures that the mapping framework is autonomous, adaptable, and able to adjust to environmental changes independently, without requiring repeated manual reconfiguration.

[0025] Another important goal of the invention is to ensure interoperability between a variety of ERP systems and cloud platforms. The system supports multiprotocol communication, including REST, SOAP, message queues, and proprietary ERP connectors, and standardizes output into cloud-compatible formats. This allows companies with different ERP ecosystems and multiple cloud providers to rely on a single, unified middleware solution instead of using fragmented connectors for each individual ERP cloud pair. The invention aims to minimize integration costs, reduce redundancies, and simplify enterprise architecture.

[0026] Another objective of the invention is to improve scalability and resilience in enterprise integration. The middleware device is designed to handle both batch-oriented data loads and near real-time transaction flows, ensuring that companies of varying sizes and industries can benefit from the solution. With a modular hardware design featuring redundant power supplies, hot-swappable communication interfaces, and edge deployment configurations, the invention aims to guarantee continuous operation even in mission-critical, high-throughput environments. By providing a resilient architecture, the invention ensures that business continuity is maintained even in the event of infrastructure failures or system overloads.

[0027] Ultimately, the invention aims to reduce the total cost of ownership for enterprise integration. By providing a hardware-based, intelligent, and adaptive system, the invention minimizes ongoing licensing costs, reduces reliance on external consultants, and lowers maintenance efforts associated with manual or semi-automated integration approaches. As a result, the invention enables companies to achieve more efficient ERP cloud integration with reduced operating costs, while simultaneously ensuring compliance, scalability, and future-proof adaptability. BRIEF DESCRIPTION OF THE FIGURE

[0028] These and other features, aspects, and advantages of the present invention will be better understood if the following detailed description is read with reference to the accompanying drawing, in which the same symbols consistently represent the same parts. The following applies: Fig. Figure 1 shows a block diagram of a system for automated, middleware-based intelligent data mapping between heterogeneous ERP systems and cloud platforms.

[0029] Experts will also recognize that the elements in the drawing are shown for the sake of simplicity and are not necessarily to scale. For example, the flowcharts illustrate the process by highlighting the main steps to enhance understanding of the aspects of this disclosure. Furthermore, with regard to the design of the device, one or more components of the device may be represented in the drawing by conventional symbols, and the drawing may show only the specific details relevant to understanding the embodiments of this disclosure, so as not to clutter the drawing with details that are readily apparent to those skilled in the art after reading this description. Detailed description of the invention

[0030] For a better understanding of the inventive principles, reference is made below to the embodiment shown in the drawing, which is described in specific terminology. However, this does not limit the scope of the invention. Changes and further modifications of the illustrated system, as well as further applications of the inventive principles, are possible, as would normally occur to a person skilled in the art in the field of invention.

[0031] It is clear to the person skilled in the art that the preceding general description and the following detailed description are exemplary and explanatory of the invention and are not intended as a limitation of it.

[0032] References in this specification to “an aspect”, “another aspect”, or similar expressions mean that a particular feature, structure, or property described in connection with the embodiment is included in at least one embodiment of the present disclosure. Therefore, occurrences of the expressions “in one embodiment”, “in another embodiment”, and similar expressions in this specification may all refer to the same embodiment, but need not.

[0033] The terms "includes," "include," or other variations thereof are intended to cover non-exclusive inclusion, such that a process or method that includes a list of steps may not only contain those steps but may also include other steps not expressly listed or inherent in such process or method. Likewise, the statement "includes..." in the case of one or more devices, subsystems, elements, structures, or components does not, without further limitations, preclude the existence of other devices, subsystems, elements, structures, components, or additional devices, subsystems, elements, structures, or components.

[0034] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as understood by a person skilled in the art in the field of the invention. The system, methods, and examples provided here serve only for illustration and are not to be construed as a limitation.

[0035] Embodiments of the present disclosure are described in detail below with reference to the attached drawing.

[0036] In Fig.Figure 1 shows a block diagram of a system for automated, middleware-based intelligent data mapping between heterogeneous ERP systems and cloud platforms. The system 100 comprises: an ERP data acquisition unit (102) with multiple physical communication ports (Ethernet, RS-485, and serial ports), a network interface controller physically connected to the communication ports, and a DMA (Direct Memory Access) controller configured to transfer raw ERP data packets into a volatile buffer without processor intervention;a semantic mapping unit (104) with an FPGA (Field Programmable Gate Array) configured as a lexical analyzer for parsing ERP schema definitions and extracting entity descriptors, and a tensor processing unit (TPU) configured to evaluate semantic similarities between heterogeneous schema elements using ontological models stored in non-volatile memory; a transformation logic unit (106) comprising a parallelized vector processor configured to perform matrix-based transformations, field normalizations, and structural reorganization of ERP data into a harmonized format, and a rule application circuit configured to perform compliance-based anonymization, unit conversions, and value scaling prior to cloud distribution;a cloud interface unit (108) comprising a protocol adapter bank with REST API encoders, SOAP translators, and message queue writers, wherein the cloud interface unit is coupled to a cryptographic controller (108a) configured with a hardware public key infrastructure (PKI) circuit to generate digital signatures and verify integrity codes for all outgoing payloads; and a feedback learning control unit (110) implemented in firmware, wherein the engine is electrically coupled to the semantic mapping unit and the transformation logic unit and is configured to monitor error logs, success rates, and acknowledgment codes received from the cloud platforms and to update the semantic rules and transformation rules stored in non-volatile memory accordingly.

[0037] In one embodiment, the ERP data acquisition unit (102) also includes a dual-port storage controller configured to allow simultaneous read / write access to multiple ERP data streams, and a timestamp circuit implemented with a real-time clock oscillator so that temporal metadata is assigned to the captured data packets for synchronization across heterogeneous ERP sources.

[0038] In one embodiment, the Semantic Mapping Unit (104) further comprises a finite state machine embedded in the FPGA, configured to tokenize scheme elements at the lexical level, and a microcode comparator circuit configured to generate weighted similarity vectors, wherein the comparator circuit transmits the vectors to the tensor processing unit for semantic evaluation.

[0039] In one embodiment, the transformation logic unit (106) comprises a multi-channel normalization subprocessor with a floating-point arithmetic kernel configured to perform currency conversions, a lookup-based coding block configured to harmonize character sets, and a masking logic circuit configured to anonymize sensitive fields by replacing them with pseudonymous identifiers derived from a cryptographic hash function.

[0040] In one embodiment, the cloud interface unit (108) also includes a multiplex transmission scheduler configured to dynamically select one of the protocol adapters based on the requirements of the cloud endpoint, and a buffer queue control circuit configured to manage payload fragmentation and retransmission while adhering to the bandwidth limitations of the cloud platform.

[0041] In one embodiment, the cryptographic controller (108a) comprises a hardware random number generator implemented with a ring oscillator circuit and a key storage physically isolated from main memory, so that session keys for encryption and authentication are generated and stored in a tamper-proof hardware enclave.

[0042] In one embodiment, the feedback learning control unit (110) further comprises a reinforcement learning circuit implemented in the firmware, wherein the circuit is configured to apply reward metrics to successful schema mappings and penalty metrics to faulty transformations, thereby adaptively changing the similarity thresholds used by the Semantic Mapping Unit.

[0043] In one embodiment, the non-volatile memory repository includes a phase-change memory array configured for the persistent storage of mapping ontologies, compliance rules, and historical error corrections, so that mapping consistency is maintained even after system restarts and power interruptions.

[0044] In one embodiment, the middleware hardware device is physically constructed as a rack-mounted enclosure that includes redundant power supplies, thermal management channels with heatpipe-assisted conduction, and hot-swappable communication interface cards, thereby ensuring continuous operation in enterprise environments under high-throughput conditions.

[0045] In one embodiment, the ERP data acquisition unit (102), the Semantic Mapping Unit, the Transformation Logic Unit, and the Cloud Interface Unit are interconnected via an internal high-speed bus selected from Peripheral Component Interconnect Express (PCIe) or Advanced eXtensible Interface (AXI), so that the latency between the units during data transformation and forwarding is minimized to less than 10 microseconds.

[0046] The present invention provides a detailed framework for a system that enables automated, middleware-based, intelligent data mapping between heterogeneous ERP systems and cloud platforms. The architecture is implemented as a dedicated middleware hardware device (MHD). The system comprises several units physically integrated via an internal high-speed bus, including a data acquisition unit, a semantic mapping unit, a transformation logic unit, a cloud interface unit, a cryptography controller, and a feedback learning control unit. All units work together in real time to enable seamless ERP-cloud integration. The system operates with a number of hardware-based technical processes that ensure accurate schema interpretation, semantic alignment, high-speed transformation, and secure data packet transmission.

[0047] The technology begins with the operation of the ERP data acquisition unit, which is configured for physical connection to heterogeneous ERP systems via multiple communication interfaces such as Ethernet, RS-485, or industrial serial ports. Incoming data packets are captured by a high-speed network interface controller and immediately forwarded to a DMA controller, which writes the packets directly to a volatile buffer. This avoids processor-related copy delays and ensures high-rate data transmission. At this stage, a timestamp generated by a real-time clock is added to each packet. This creates time markers that facilitate synchronization when processing multiple ERP streams simultaneously.A dual-port storage controller enables simultaneous read and write access, thus ensuring collision-free data transfer from multiple sources.

[0048] Once the data is buffered, the Semantic Mapping Unit performs the first level of schema interpretation. This unit uses an FPGA-based lexical analyzer configured as a finite state machine to tokenize incoming data streams into identifiable schema components such as field separators, table boundaries, and relational dependencies. The FPGA implements deterministic logic gates that analyze ERP data formats quickly and without software parsing overhead. The analyzed schema tokens are passed to a microcoded comparator circuit that generates weighted similarity vectors. These vectors represent the degree of structural and semantic correlation between the extracted schema tokens and the ontology descriptors stored in non-volatile memory.The comparator circuit transmits the similarity vectors to a Tensor Processing Unit (TPU), which applies machine learning techniques to determine the semantic equivalence of ERP fields. The TPU uses deep-embedding models trained on historical ERP cloud mappings. This allows it to identify fields that share a common semantic meaning despite syntactic differences, such as "Receipt_ID" in one ERP system corresponding to "StaftNumber" in another.

[0049] The technology then transitions to the Transformation Logic Unit, which performs normalization, structural reorganization, and compliance-based adjustments. At its core, this unit uses a parallelized vector processor that performs matrix-based transformations of the data. For example, nested hierarchical objects in ERP records are converted into normalized relational rows, as required by most cloud platforms. Similarly, composite fields such as "CustomerName" are decomposed into atomic units like "FirstName" and "LastName" according to the rules of the target schema. A floating-point arithmetic kernel performs unit conversions in real time, including adjustments to currency, weight, or volume units, based on lookup tables stored in the repository. Dedicated masking logic anonymizes sensitive identifiers using a cryptographic hash function that generates pseudonymous values.The compliance verification subprocessor then compares the transformed fields with regulatory metadata and ensures that field lengths, permitted formats, and anonymization policies comply with requirements such as GDPR or HIPAA. Only after the compliance processor validates the transformations does the data package reach the Cloud Interface Unit.

[0050] The Cloud Interface Unit performs the encapsulation and delivery techniques. A multiplex transmission scheduler selects the appropriate protocol adapter—REST, SOAP, or message queue—based on metadata describing the destination cloud endpoint. Each payload passes through a buffer queue controller, which fragments or aggregates packets based on bandwidth requirements to ensure optimal throughput without congestion. A serialization-deserialization module (SERDES) re-encodes the payload into binary or JSON-based structures, as required by the Cloud Ingestion API. Before transmission, the payload is processed by the cryptographic controller, which uses a hardware random number generator to create a session key and signs the payload with a PKI-based digital signature.A hardware-isolated vault stores private keys, ensuring that cryptographic integrity is maintained even if main memory is compromised. The signed payload is then transmitted to the cloud endpoint, where the integrity codes are verified.

[0051] A critical component of the system is the Feedback Learning Control Unit, which continuously refines mappings and transformations. After the payload is sent, acknowledgment codes, error reports, and latency metrics are collected from the cloud platforms and fed back to the firmware-based learning engine. Within this engine, a reinforcement learning technique applies reward metrics when the schema mappings are correct and penalties when mismatches or compliance violations occur. These results adjust the similarity thresholds used by the comparator circuit and the TPU, thus optimizing the semantic mapping technique for subsequent iterations. Additionally, the telemetry collector within the engine monitors bus conflicts, processor utilization, and retransmission rates, redistributing workloads between the FPGA and TPU to balance performance.For example, if an overload of the FPGA is detected during lexical parsing, certain preprocessing functions can be offloaded to the TPU to ensure consistent throughput across the entire pipeline.

[0052] The non-volatile memory repository acts as the system's persistent knowledge base. Based on phase-change memory arrays, it stores ontology models, historical bug fixes, regulatory metadata, and transformation rules. Thanks to its non-volatile nature, the repository retains mappings even after restarts and power outages, ensuring seamless system restoration of its knowledge base. During runtime, the Semantic Mapping Unit, the Transformation Logic Unit, and the Feedback Learning Control Unit access this repository concurrently, enabling shared intelligence across these units.

[0053] The middleware hardware device is physically housed in a rack-mounted enclosure that includes redundant power supplies and a heatpipe-based thermal management system to prevent performance degradation under high throughput. The enclosure also contains hot-swappable communication interface cards, allowing organizations to update or replace communication modules without interrupting operations. An integrated Trusted Platform Module (TPM) enforces secure boot procedures and ensures that only authenticated firmware images are loaded to initialize the FPGA and TPU. This guarantees that malicious firmware cannot modify the semantic mapping logic and provides a robust security foundation.

[0054] The system's overall technology can be described as a continuous pipeline: ingestion, analysis, semantic evaluation, transformation, conformance check, encapsulation, cryptographic signing, transmission, and adaptive feedback. Each step is hardware-driven, reducing latency and avoiding bottlenecks common in pure software middleware. By combining deterministic FPGA parsing with probabilistic TPU-based semantic evaluation, the process achieves both speed and adaptive intelligence. The continuous refinement loop ensures that the system automatically updates its mappings as ERP schemas and cloud platforms evolve, eliminating the need for manual reconfiguration or consultant intervention.

[0055] For practical application, consider an example where a company uses SAP ERP for finance and Microsoft Dynamics ERP for supply chain management, storing consolidated data on a cloud-based analytics platform. The system ingests financial records with currency values ​​and supply chain records with inventory quantities. The Semantic Mapping Unit maps fields such as "Ledger_Currency" and "CurrencyCode," as well as "Inventory_Units" and "Stock_Qty." The Transformation Logic Unit converts currency values ​​into a standardized format and transforms inventory units into the base measurement defined by the cloud schema. The Compliance Processor verifies the anonymization of customer identifiers before the Cloud Interface Unit sends the data via a REST API encoder. The Feedback Learning Control Unit monitors whether the cloud confirms correct ingestion and updates the ontology when new fields appear in future SAP upgrades.This technical cycle repeats continuously, offering robust, adaptive, and rapid integration.

[0056] Thus, the invention offers not only a structural, hardware-based device, but also a complete technical framework in which ERP data undergoes a pipeline of semantic interpretation, rule-based transformation, conformance checking, secure encapsulation, and adaptive feedback. By embedding these processes in specialized hardware units and firmware circuits, the invention achieves a level of speed, accuracy, adaptability, and security that overcomes the limitations of existing ERP cloud integration solutions.

[0057] The invention will now be described in more detail with regard to its structural and functional components. The Middleware Hardware Device (MHD) is implemented as a rack-mountable device and comprises multi-core processors, FPGA logic blocks, high-speed network interface controllers, and secure memory modules.

[0058] The ERP data acquisition unit consists of dual-port Ethernet controllers, RS-485 industrial connectors, and a DMA (Direct Memory Access) controller that simultaneously ingests ERP data from multiple sources. This unit performs initial packet framing, timestamping, and buffer queueing before forwarding the packets to the Semantic Mapping Unit.

[0059] The Semantic Mapping Unit is implemented through a hybrid architecture that combines hardware accelerators and software-defined processors. FPGA-based lexical analyzers operate as finite automata, analyzing ERP schema definitions, identifying field boundaries, and recognizing ontology descriptors. A machine learning co-processor embedded as a Tensor Processing Unit (TPU) evaluates semantic similarities between heterogeneous schema elements and dynamically updates the mapping rules stored in the local knowledge base.

[0060] The Transformation Logic Unit performs the structural reorganization of ERP data using hardware-optimized transformation matrices. For example, nested data structures from an ERP finance module are transformed into normalized relational structures suitable for cloud deployment. Complex field transformations, such as unit conversions, currency normalization, or compliance-based anonymization, are handled by vectorized processing kernels that execute the transformations in parallel.

[0061] The Cloud Interface Unit contains a protocol adapter bank with REST API encoders, SOAP translators, and message queue writers (e.g., AMQP, Kafka). Depending on the target cloud platform, the interface automatically selects the appropriate transmission protocol. Before transmission, the cryptography controller adds integrity codes and digital signatures using a hardware-based public key infrastructure (PKI).

[0062] The invention also includes a feedback learning control unit implemented in the firmware, which monitors mapping success rates, error logs, and cloud-side confirmation codes. This feedback loop automatically refines the semantic matching techniques and transformation rules stored in the repository, thus enabling continuous adaptation to evolving ERP and cloud schemas.

[0063] Physically, the MHD is housed in a shielded enclosure with thermal management channels, redundant power supplies, and hot-swappable communication cards. The device can be deployed as a middleware node on-premises or at the network edge, connecting multiple ERP systems to cloud backbones.

[0064] The present invention relates to the integration of enterprise-wide information technologies, and in particular systems and devices for intelligent data mapping based on middleware. The invention addresses the technical challenge of harmonizing heterogeneous ERP data models into standardized formats usable by cloud platforms. This is achieved through the use of a dedicated hardware architecture that combines data acquisition, semantic interpretation, structural transformation, compliance enforcement, and secure protocol adaptation. The field encompasses enterprise middleware systems, ERP interoperability frameworks, cloud data acquisition devices, and hardware-accelerated semantic mapping technologies, with a focus on adaptive learning, high-speed data processing, and cryptographic security in enterprise-to-cloud integration.

[0065] The drawing and the preceding description show examples of embodiments. Those skilled in the art will recognize that one or more of the described elements can be combined to form a single functional element. Alternatively, certain elements can be divided into several functional elements. Elements of one embodiment can be added to another embodiment. For example, the sequence of the processes described here can be changed and is not limited to the manner described here. Furthermore, the actions of a flowchart need not be implemented in the sequence shown; nor does it necessarily have to be performed by all actions. Actions that are not dependent on other actions can also be performed in parallel with the other actions. The scope of the embodiments is in no way limited by these specific examples.Numerous variations are possible, whether explicitly stated in the specification or not, such as differences in structure, dimensions, and material use. The range of embodiments is at least as broad as specified in the following claims.

[0066] Advantages, further benefits, and problem solutions have been described above with reference to specific embodiments. However, the advantages, benefits, problem solutions, and all components that can lead to an advantage, benefit, or solution occurring or becoming more apparent are not to be construed as critical, necessary, or essential features or components of individual or all claims. REFERENCES 100 A system for automated, middleware-based intelligent data mapping between heterogeneous ERP systems and cloud platforms. 102 ERP data capture unit 104 Semantic mapping unit 106 Transformation logic unit 108 Cloud interface unit 110 Feedback Learning Control Unit 112 Cryptography Controllers

Claims

[1] A system for automated, middleware-based intelligent data mapping between heterogeneous enterprise resource planning (ERP) systems and cloud platforms, comprising the following: an ERP data acquisition unit with a variety of physical communication ports, selected from Ethernet and serial ports, a network interface controller physically connected to the communication ports, and a DMA (Direct Memory Access) controller configured to transfer raw ERP data packets into a volatile buffer memory without processor intervention; a Semantic Mapping Unit consisting of a Field Programmable Gate Array (FPGA) configured as a lexical analyzer to analyze ERP schema definitions and extract entity descriptors, and a Tensor Processing Unit (TPU) configured to evaluate the semantic similarity between heterogeneous schema elements based on ontological models stored in a non-volatile memory repository; a transformation logic unit comprising a parallelized vector processor configured to perform matrix-based transformations, field normalizations and structural reorganization of ERP data into a harmonized format, and a rule application circuit configured to perform compliance-based anonymization, unit conversions and value scaling prior to cloud distribution; a cloud interface unit comprising a protocol adapter bank with encoders, translators, and message queue writers, wherein the cloud interface unit is coupled with a cryptographic controller configured with a hardware public key infrastructure (PKI) circuit to generate digital signatures and verify integrity codes for all outgoing payloads; and a Feedback Learning Control Unit implemented in the firmware, wherein the engine is electrically coupled to the Semantic Mapping Unit and the Transformation Logic Unit and is configured to monitor error logs, success rates and acknowledgment codes received from the cloud platforms and to update the semantic and transformation rules stored in the non-volatile memory repository accordingly. [2] System according to claim 1, wherein the ERP data acquisition unit further comprises a dual-port storage controller configured to enable simultaneous read / write access to multiple ERP data streams, and a timestamp circuit implemented with a real-time clock oscillator, so that temporal metadata is assigned to the captured data packets for synchronization across heterogeneous ERP sources. [3] System according to claim 1, wherein the Semantic Mapping Unit further comprises a finite state machine embedded in the FPGA, configured to tokenize scheme elements at the lexical level, and a microcoded comparator circuit configured to generate weighted similarity vectors, wherein the comparator circuit transmits the vectors to the tensor processing unit for semantic evaluation. [4] System according to claim 1, wherein the transformation logic unit comprises a multi-channel normalization subprocessor comprising a floating-point arithmetic kernel configured to perform currency conversions, a lookup-based encoding block configured to harmonize character sets, and a masking logic circuit configured to anonymize sensitive fields by replacing them with pseudonymous identifiers derived from a cryptographic hash function. [5] System according to claim 1, wherein the cloud interface unit further comprises a multiplex transmission scheduler configured to dynamically select one of the protocol adapters based on the requirements of the cloud endpoint, and a buffer queue control circuit configured to manage payload fragmentation and retransmission in accordance with the bandwidth limitations of the cloud platform. [6] System according to claim 1, wherein the cryptographic controller comprises a hardware random number generator implemented with a ring oscillator circuit and a key storage physically isolated from main memory, such that session keys for encryption and authentication are generated and stored in a tamper-proof hardware enclave. [7] System according to claim 1, wherein the feedback learning control unit further comprises a reinforcement learning circuit implemented in the firmware, the circuit being configured to apply reward metrics to successful schema mappings and penalty metrics to faulty transformations, thereby adaptively changing the similarity thresholds used by the Semantic Mapping Unit. [8] System according to claim 1, wherein the non-volatile memory repository comprises a phase-change memory array configured for the persistent storage of mapping ontologies, conformance rules and historical error corrections, so that mapping consistency is maintained even during system restarts and power interruptions. [9] System according to claim 1, wherein the middleware hardware device is physically constructed as a rack-mounted enclosure comprising redundant power supplies, thermal management channels with heat pipe-assisted conduction and hot-swappable communication interface cards, thereby ensuring continuous operation in enterprise environments under high-throughput conditions.

Images