Design And Verification Of Communication Middleware For Embedded Computing Platform

As a kind of basic supporting software,communication middleware is mainly used in IP network to solve the problem of interaction between applications,which are in different systems.With the diversification of data processing business scenarios,a dedicated embedded computing platform composed of specific processors is becoming one of the main schemes for system deployment.Heterogeneous systems can give full play to the characteristics of each processor and efficiently solve actual business problems,But the high degree of heterogeneity also increases the complexity of the software and hardware of the system,which poses a challenge to the rapid integration of distributed applications.Therefore,it is necessary to design a communication middleware suitable for embedded computing platforms.Based on the actual engineering requirements,this paper studies and analyzes the data transmission,application interaction across communication middleware and communication optimization for resource-limited devices,and puts forward corresponding solutions.1.A data transmission scheme based on link label and suitable for heterogeneous network environment is proposed.In the embedded computing platform,the heterogeneous network leads to the complex problem of data transmission and forwarding.a data transmission scheme based on link labels and suitable for heterogeneous network environments is proposed.This solution draws on the design architecture of SDN,separates the management and forwarding functions in the network,and makes the network more flexible.In order to realize the data transmission in heterogeneous networks,the source routing scheme in SR-MPLS is adopted.The heterogeneous network is first modeled,and each communication link in the network is assigned a unique label.Firstly,the heterogeneous network is modeled,and a unique label is assigned to each communication link in the network.Then,the transmission path information composed of link label list is embedded in the data packet to guide the data forwarding.Compared with the traditional forwarding scheme based on destination address lookup table,this scheme is more flexible and can provide more refined communication services for different streams.2.An embedded cross-communication middleware application interaction scheme is proposed.Due to the lack of interoperability between different communication middleware,different platform integration needs the help of gateway or other thirdparty network entities,which results in poor flexibility and universality of the system,a lightweight embedded cross communication middleware interaction scheme is proposed.The protocol conversion function is transferred from the external network entities such as gateway to the internal of general communication middleware,and the consistency of the original API is guaranteed to the greatest extent.The interaction between the applications of cross communication middleware can be realized under the condition of modifying a small amount of upper application code.In this paper,MQTT and DDS communication middleware are taken as the research object,the differences of protocol interoperability,syntax interoperability and semantic interoperability between the two objects are analyzed,and a topic mapping mechanism based on component model is proposed to realize the semantic interoperability between them,an intermediate transformation scheme based on JSON file is proposed to realize the syntax interoperability between them,which provides a reference for the interaction between different communication middleware.3.A symmetric data transceiver scheme based on Kalman filter model is proposed.In the embedded computing platform,the bandwidth and energy of the sensor devices is limited,and the channel environment is unstable,and continuous and accurate data cannot be provided to other devices,which makes the system unable to make corresponding actions based on the actual environment information in time,resulting in insufficient system reliability,and a symmetric data transceiver scheme based on Kalman filter model is proposed.This solution deploys the same prediction model on the sender and receiver.The sender determines whether the data is sent by comparing whether the difference between the observed value and the local predicted value exceeds the fault tolerance threshold.when the receiver receive no data,the predicted value will be used as the sensor data from the sender,which can reduce bandwidth occupation and energy consumption while tolerating certain data errors.