Vehicle-mounted high-performance computing framework based on SOA communication middleware and IWARP technology and implementation method

Abstract

The application discloses a kind of based on SOA communication middleware and IWARP technology's vehicle-mounted high-performance computing framework and implementation method, including central computing unit, edge computing unit and domain controller, communicate with each other by SOA communication middleware;Central computing unit and edge computing unit built-in network controller RNIC, send and receive iWARP protocol message, central computing unit and multiple edge computing units form distributed computing architecture;Domain controller sends to central computing unit after data acquisition, central computing unit will be split after data transmission to edge computing unit and carry out distributed computing, then receive the calculation result returned and carry out data fusion, obtain final result and send to corresponding domain controller.The application adds the RNIC required by iWARP technology between central computing unit and edge computing unit, simultaneously using SOA communication middleware, constructs a set of high-availability, high-performance, low-latency distributed computing framework, to guarantee the faster implementation of automatic driving application scenario technology level.

Description

Technical Field

[0001] This invention relates to autonomous driving technology, and more particularly to an in-vehicle high-performance computing framework and implementation method based on SOA communication middleware and IWARP technology. Background Technology

[0002] With the continuous evolution of automotive Ethernet technology and the increasing centralization of next-generation automotive electronic and electrical architectures, domain-based hardware architectures are becoming increasingly prevalent. At the software architecture level, Service-Oriented Architecture (SOA) is a necessary condition for realizing a centralized automotive E / E architecture. In-vehicle communication middleware, such as DDS, SOME / IP, and MQTT, all support SOA programming models. However, these three types of in-vehicle communication middleware run on the TCP / IP protocol stack. The back-and-forth data transfer between system memory, processor cache, and network controller cache places a heavy burden on the processor and memory, further exacerbating network latency. Excessive latency is unacceptable for most in-vehicle applications, such as autonomous driving.

[0003] Meanwhile, the realization of medium- and high-level standards under the autonomous driving classification system depends on the development of computing power in MPU chips. This means that once chip computing power encounters a bottleneck and cannot be improved, it will in turn restrict the technological level of autonomous driving. Therefore, designing a distributed in-vehicle high-performance computing framework is particularly important. The technology of implementing a distributed high-performance computing framework in an in-vehicle environment based on SOA software architecture communication middleware and IWARP technology effectively fills the gaps mentioned above. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide an in-vehicle high-performance computing framework and implementation method based on SOA communication middleware and IWARP technology. This involves deploying a network controller RNIC supporting IWARP technology within the controller or computing unit corresponding to the MCU / MPU, and utilizing communication middleware supporting SOA software architecture to realize an in-vehicle distributed high-performance computing framework.

[0005] To achieve the objectives of this invention, the technical solution adopted is as follows:

[0006] A vehicle-mounted high-performance computing framework based on SOA communication middleware and IWARP technology includes a central computing unit, an edge computing unit, and a domain controller.

[0007] The central computing unit and the edge computing unit communicate with each other through SOA communication middleware, and the central computing unit and the domain controller communicate with each other through SOA communication middleware;

[0008] The central computing unit has a built-in network controller RNIC, and the edge computing units have built-in network controller RNICs to send and receive iWARP protocol messages. The central computing unit and multiple edge computing units form a distributed computing architecture.

[0009] After collecting data, the domain controller sends it to the central computing unit. The central computing unit then segments the data and sends it to the edge computing unit for distributed computing. The unit then receives the returned computing results, fuses the data, and obtains the final result, which is then sent to the corresponding domain controller.

[0010] Furthermore, domain controllers communicate with each other through SOA communication middleware.

[0011] Furthermore, the SOA communication middleware is one of SOME / IP, DDS, or MQTT.

[0012] Furthermore, the SOA communication middleware is adapted to iWARP technology, enabling communication between the central computing unit and the edge computing unit through the SOA communication middleware that supports iWARP technology.

[0013] Furthermore, the central computing unit and the edge computing units communicate via a Request / Response model; specifically:

[0014] The central computing unit encapsulates the segmented data as a data source in a Request message and sends the Request to the edge computing unit. After receiving the Request message, the edge computing unit extracts the data source, calls its own computing power and corresponding algorithms to perform calculations, and outputs the calculation results.

[0015] The edge computing unit encapsulates the calculation results as a data source in a Response message and sends the Response message to the central computing unit. After receiving the Response message, the central computing unit extracts the calculation results and, together with the calculation results returned by other edge computing units, obtains a final result through data fusion.

[0016] Furthermore, the central computing unit and the domain controller communicate through two types of communication models: the Notification model provided in SOA, which allows for unidirectional data flow; or the Request / Response model provided in SOA, which allows for bidirectional data flow.

[0017] Furthermore, the domain controller collects data and sends it to the central computing unit through the Notification mode of the SOA communication middleware; after the central computing unit calculates the data through the distributed high-performance computing framework, it sends the calculation result back to the domain controller through the Notification mode of the SOA communication middleware.

[0018] Furthermore, the domain controller collects data and sends it to the central computing unit through the Notification mode of the SOA communication middleware; after the central computing unit calculates the data through the distributed high-performance computing framework, it sends the calculation result to the domain controller through the Request mode of the SOA communication middleware.

[0019] After receiving the request, the domain controller extracts the calculation results and performs secondary analysis. Then, the analysis results are sent to the central processing unit through the Response mode of the SOA communication middleware.

[0020] An implementation method for an in-vehicle high-performance computing framework based on SOA communication middleware and IWARP technology includes the following steps:

[0021] 1) After the domain controller collects data, it sends the collected data to the central processing unit;

[0022] 2) The central processing unit divides the data according to the total number of edge computing units and assigns them sequential numbers. At the same time, a timer is started to avoid the impact of the offline edge computing units on data calculation.

[0023] 3) The central processing unit encapsulates the corresponding data according to the sequence number and sends it to the edge computing unit through the Request mode;

[0024] 4) After the edge computing unit obtains the data, it calls its own resident algorithm module to perform data calculations and outputs the calculation results;

[0025] 5) The edge computing unit encapsulates the sequence number and calculation result provided in the Request message into a Response message and sends the message to the central computing unit.

[0026] 6) The central processing unit receives the Response message provided by the corresponding edge computing unit, extracts the sequence number and calculation result, and continues to wait for other edge computing units to submit corresponding calculation results until the timer expires; if the timer expires, the calculation process should return failure.

[0027] 7) The central processing unit determines whether it has received response messages from all edge computing units. If all have been received, it proceeds to process 8); if not all have been received, it returns to process 6.

[0028] 8) The central computing unit merges all the collected calculation results to obtain the final result and sends it to the corresponding domain controller.

[0029] The beneficial effects of this invention are that, compared with the prior art, this invention utilizes a centralized E / E domain architecture, adds an RNIC required by iWARP technology between the vehicle central computing unit and the vehicle edge computing unit, and utilizes the Request / Response communication mode and Notification communication mode commonly provided by SOA communication middleware to build a highly available, high-performance, and low-latency vehicle distributed computing framework.

[0030] This distributed computing framework effectively enhances the overall computing power of the vehicle by increasing the number of chips rather than increasing chip computing power, thereby ensuring the faster realization of the technical level of autonomous driving application scenarios.

[0031] Meanwhile, leveraging the Internet's wide-area RDMA protocol technology, it effectively meets the stringent latency requirements of certain application scenarios in the automotive field. iWARP technology allows for rapid integration into automotive Ethernet, rather than replacing it, essentially providing a low-cost solution.

[0032] This invention also utilizes the characteristics of SOA to enable the in-vehicle distributed computing framework to ensure high reliability and low latency in application scenarios and environments such as emergency braking, ADAS, and smart cockpit, thereby ensuring the safety and comfort of the entire vehicle during operation. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of the in-vehicle high-performance computing framework based on SOA communication middleware and IWARP technology described in this invention;

[0034] Figure 2 This is a schematic diagram of SOA communication interaction between the central computing unit and the edge computing unit;

[0035] Figure 3 This is a schematic diagram of SOA communication interaction between the central processing unit and the domain controller;

[0036] Figure 4 It is the specific execution flow of a distributed computing framework. Detailed Implementation

[0037] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of this application.

[0038] RDMA is a remote direct memory access technology that allows computers to directly read each other's memory without processor processing, significantly reducing the overhead of processor data copying. The act of copying data is a major factor contributing to increased latency.

[0039] Internet Wide Area RDMA Protocol (iWARP) is RDMA over TCP, allowing RDMA network protocols to be executed over the TCP / IP protocol stack. Unlike other RDMA technologies, iWARP only requires the network interface card (NIC) to be an iWARP-enabled RNIC, allowing RDMA to be used on standard Ethernet infrastructure (switches). iWARP technology can be quickly integrated into automotive Ethernet systems, rather than replacing them, essentially providing a low-cost solution.

[0040] like Figure 1 The diagram shows a typical automotive E / E architecture hardware structure. Domain controller A communicates with domain controllers B and C via SOA communication middleware. SOA communication middleware refers to one of three technologies: SOME / IP, DDS, or MQTT. SOA communication middleware is based on the IP protocol, regardless of whether the transport protocol over it is TCP or UDP.

[0041] The central computing unit has a built-in network controller (RNIC) for sending and receiving iWARP protocol messages. The central computing unit communicates with domain controllers A, B, and C via SOA middleware. Together with four edge computing units, the central computing unit forms a distributed computing architecture. Each edge computing unit has a built-in RNIC for sending and receiving iWARP protocol messages.

[0042] SOA communication middleware needs to be compatible with iWARP technology. Based on this, the central computing unit and the edge computing unit communicate with each other through SOA communication middleware that supports iWARP technology.

[0043] like Figure 2 As shown, the central computing unit and the edge computing unit communicate via a Request / Response model. Specifically, 1) the central computing unit encapsulates the segmented data as a data source in a Request message and sends a Request to the edge computing unit; 2) upon receiving the Request message, the edge computing unit extracts the data source, utilizes its own MPU's computing power and corresponding algorithms to perform calculations, and then outputs the calculation results; 3) the edge computing unit encapsulates the calculation results as a data source in a Response message and sends a Response message to the central computing unit; 4) upon receiving the Response message, the central computing unit extracts the calculation results and, together with the calculation results returned by other edge computing units, performs data fusion to obtain a final result. This final result is then used as the data source to communicate with the domain controller via Notification mode or Request / Response mode.

[0044] Therefore, edge computing units need to host algorithms, while central computing units provide data.

[0045] like Figure 3 As shown, the central computing unit and the domain controller communicate through two communication models. The first is the Notification model provided in SOA, where data flows unidirectionally. The second is the Request / Response model provided in SOA, where data flows bidirectionally. Figure 3 A typical case is provided, specifically involving three scenarios:

[0046] Scenario 1: Domain controller C collects data, such as data from LiDAR and camera video, and sends the data to the central processing unit (CPU) via the Notification mode of the SOA communication middleware. The CPU then processes the data using a distributed high-performance computing framework and sends the results back to domain controller C via the Notification mode of the SOA communication middleware. This scenario typically occurs in ADAS applications.

[0047] Scenario 2: Domain controller C collects data, such as data from LiDAR and camera video, and sends the data to the central processing unit (CPU) via the Notification mode of the SOA communication middleware. The CPU then calculates the data using a distributed high-performance computing framework and sends the calculation result back to domain controller A via the Notification mode of the SOA communication middleware. This scenario typically occurs when domain controller A, acting as the vehicle body domain or powertrain domain, needs to perform emergency braking.

[0048] Scenario 3: Domain controller C collects data, such as from LiDAR and camera video, and sends the data to the central processing unit (CPU) via the Notification mode of the SOA communication middleware. The CPU then processes the data using a distributed high-performance computing framework and sends the results to domain controller B via the Request mode of the SOA communication middleware. Upon receiving the request, domain controller B extracts the results, performs secondary analysis, and then sends the analysis results back to the CPU via the Response mode of the SOA communication middleware. This scenario typically occurs when domain controller B acts as the cockpit domain and needs to provide feedback on relevant cockpit and instrument information.

[0049] The following is a brief description of the specific execution flow of a distributed computing framework. For example... Figure 4 As shown.

[0050] 1) After the domain controller or other vehicle-mounted unit collects data, it sends the collected data to the central processing unit;

[0051] 2) The central processing unit divides the data according to the total number of edge computing units and assigns them sequential numbers. At the same time, a timer is started to avoid the impact of the offline edge computing units on data calculation.

[0052] 3) The central processing unit encapsulates the corresponding data according to the sequence number and sends it to the edge computing unit through the Request mode;

[0053] 4) After the edge computing unit obtains the data, it calls its own resident algorithm module to perform data calculations and outputs the calculation results;

[0054] 5) The edge computing unit encapsulates the sequence number and calculation result provided in the Request message into a Response message and sends the message to the central computing unit.

[0055] 6) The central processing unit receives the Response message from the corresponding edge computing unit, extracts the sequence number and calculation result, and continues to wait for other edge computing units to submit their corresponding calculation results until the timer expires. If the timer expires, the calculation process should return failure.

[0056] 7) The central processing unit determines whether it has received response messages from all edge computing units. If all have been received, it proceeds to step 8); otherwise, it returns to step 6.

[0057] 8) The central computing unit merges all the collected calculation results to obtain the final result and sends it to the corresponding domain controller.

[0058] In summary, through Figure 1 , Figure 2 , Figure 3 This paper describes an in-vehicle high-performance computing framework based on SOA communication middleware and IWARP technology. Figure 4 The implementation method of an in-vehicle high-performance computing framework based on SOA communication middleware and IWARP technology is described.

[0059] The beneficial effects of this invention are that, compared with the prior art, this invention utilizes a centralized E / E domain architecture, adds an RNIC required by iWARP technology between the vehicle central computing unit and the vehicle edge computing unit, and utilizes the Request / Response communication mode and Notification communication mode commonly provided by SOA communication middleware to build a highly available, high-performance, and low-latency vehicle distributed computing framework.

[0060] This distributed computing framework effectively enhances the overall computing power of the vehicle by increasing the number of chips rather than increasing chip computing power, thereby ensuring the faster realization of the technical level of autonomous driving application scenarios.

[0061] Meanwhile, leveraging the Internet's wide-area RDMA protocol technology, it effectively meets the stringent latency requirements of certain application scenarios in the automotive field. iWARP technology allows for rapid integration into automotive Ethernet, rather than replacing it, essentially providing a low-cost solution.

[0062] This invention also utilizes the characteristics of SOA to enable the in-vehicle distributed computing framework to ensure high reliability and low latency in application scenarios and environments such as emergency braking, ADAS, and smart cockpit, thereby ensuring the safety and comfort of the entire vehicle during operation.

[0063] The applicant of this invention has provided a detailed description of the embodiments of the invention in conjunction with the accompanying drawings. However, those skilled in the art should understand that the above embodiments are merely preferred embodiments of the invention. The detailed description is only intended to help readers better understand the spirit of the invention and is not intended to limit the scope of protection of the invention. On the contrary, any improvements or modifications made based on the inventive spirit of the invention should fall within the scope of protection of the invention.

Claims

1. A vehicle-mounted high-performance computing system based on SOA communication middleware and IWARP technology, characterized in that, This includes a central computing unit, edge computing units, and domain controllers; The central computing unit and the edge computing unit communicate with each other through SOA communication middleware that supports iWARP technology, and the central computing unit and the domain controller communicate with each other through SOA communication middleware. The central computing unit has a built-in network controller RNIC, and the edge computing units have built-in network controller RNICs to send and receive iWARP protocol messages. The central computing unit and multiple edge computing units form a distributed computing architecture. After the domain controller collects data, it sends it to the central computing unit. The central computing unit then segments the data and sends it to the edge computing unit for distributed computing. It then receives the returned computing results, fuses the data, obtains the final result, and sends it to the corresponding domain controller. The central computing unit and the edge computing units communicate via a Request / Response model; specifically: The central processing unit divides the data according to the total number of edge computing units and assigns them sequential numbers, while simultaneously starting a timer; The central processing unit encapsulates the segmented data as a data source in a Request message according to the sequence number and sends the Request message to the edge computing unit. After receiving the Request message, the edge computing unit extracts the data source, calls its own computing power and corresponding algorithms to perform calculations, and outputs the calculation results; The edge computing unit encapsulates the calculation result as a data source in the Response message based on the sequence number provided in the Request message, and sends the Response message to the central computing unit. After receiving the Response message, the central computing unit extracts the calculation result and sequence number, and continues to wait for other edge computing units to submit their corresponding calculation results until the timer expires. The central computing unit determines whether it has received Response messages from all edge computing units. If it has received all of them, it combines the calculation results returned by all edge computing units and obtains a final result through data fusion.

2. The in-vehicle high-performance computing system based on SOA communication middleware and IWARP technology according to claim 1, characterized in that, Domain controllers communicate with each other through SOA communication middleware.

3. The in-vehicle high-performance computing system based on SOA communication middleware and IWARP technology according to claim 1, characterized in that, SOA communication middleware is one of three: SOME / IP, DDS, or MQTT.

4. The in-vehicle high-performance computing system based on SOA communication middleware and IWARP technology according to claim 1, characterized in that, The central computing unit and the domain controller communicate through two types of communication models: the Notification model provided in SOA, which allows for unidirectional data flow; or the Request / Response model provided in SOA, which allows for bidirectional data flow.

5. The in-vehicle high-performance computing system based on SOA communication middleware and IWARP technology according to claim 4, characterized in that, The domain controller collects data and sends it to the central computing unit via the Notification mode of the SOA communication middleware. After the central computing unit calculates the data using a distributed high-performance computing framework, it sends the calculation results back to the domain controller via the Notification mode of the SOA communication middleware.

6. The in-vehicle high-performance computing system based on SOA communication middleware and IWARP technology according to claim 4, characterized in that, The domain controller collects data and sends it to the central processing unit through the Notification mode of the SOA communication middleware. After the central processing unit calculates the data through the distributed high-performance computing framework, it sends the calculation result back to the domain controller through the Request mode of the SOA communication middleware. After receiving the request, the domain controller extracts the calculation results and performs secondary analysis. Then, the analysis results are sent to the central processing unit through the Response mode of the SOA communication middleware.

7. A vehicle-mounted high-performance computing method based on SOA communication middleware and IWARP technology, based on the vehicle-mounted high-performance computing system based on SOA communication middleware and IWARP technology as described in any one of claims 1-6, characterized in that, Including the following steps: 1) After the domain controller collects data, it sends the collected data to the central processing unit; 2) The central processing unit divides the data according to the total number of edge computing units and assigns them sequential numbers, while simultaneously starting a timer; 3) The central processing unit encapsulates the corresponding data according to the sequence number and sends it to the edge computing unit through a Request message; 4) After acquiring the data, the edge computing unit uses its own computing power and corresponding algorithms to perform data calculations and outputs the calculation results; 5) The edge computing unit encapsulates the sequence number and calculation result provided in the Request message into a Response message and sends the message to the central computing unit. 6) The central processing unit receives the Response message provided by the corresponding edge computing unit, extracts the sequence number and calculation result, and continues to wait for other edge computing units to submit corresponding calculation results until the timer expires; if the timer expires, the calculation process should return failure. 7) The central processing unit determines whether it has received response messages from all edge computing units. If all have been received, proceed to step 8; otherwise, return to step 6. 8) The central computing unit merges all the collected calculation results to obtain the final result and sends it to the corresponding domain controller.

Images