A radar software cross-platform computing system

Abstract

The present application relates to a kind of radar software cross-platform computing techniques, including the abstraction of radar information processing software component unit into vector computing unit, parallel computing unit, memory space allocation unit, system function calling unit, data type and size end conversion unit, each unit is adapted to the heterogeneity of each platform, and provides the function calling interface of generalization.The radar information processing software cross-platform computing technique of the present application mainly contains vector computing unit, parallel computing unit, memory address allocation unit, system function calling unit, data type and size end conversion unit, adopts the generalization interface to abstract radar information processing software component unit, supports the cross-platform transplantation of radar information processing software, so that radar information processing software can be transplanted across platform after once development of function.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of radar information processing, in particular to a radar software cross-platform computing technology. BACKGROUND

[0002] With the rapid development of new weapons such as missiles, stealth aircraft, stealth warships and near-space hypersonic vehicles, China will face air and space target threats in the future, including all airspace, multi-dimensional and all-weather threats. The main threats to radars include new generation fighter jets, new generation bombers, unmanned aerial vehicles, hypersonic strike weapons, kinetic energy strike weapons, directional energy strike weapons, network attacks, etc. The battlefield environment of the radar includes complex geographical environment, weather environment, complex electromagnetic interference environment, etc. The diversification and complexity of radar combat tasks require the rapid application of new functions and new algorithms to quickly respond to new military threats.

[0003] At present, the radar development process is directional development, and the characteristics of one-time, one-type, and narrow application range are very prominent. Moreover, the equipment type architecture is closed, the application development threshold is high, the technology updates slowly, the compatibility and universality are poor, the maintenance and support are complex, and the position upgrading and modification are difficult.

[0004] The traditional radar software development process is to verify the algorithm function of the digital prototype, complete the real-time processing software design, coding, integration, and compilation based on the embedded integrated development environment, and download the compiled executable file to the embedded device for running and debugging. The embedded software development cycle is long, and the debugging process is complex. The embedded platform architecture is different, and the code transplantation is difficult, which leads to a very tedious process of complex embedded system software development, integration and debugging, algorithm upgrade difficulty, and long time consumption. Therefore, a radar software cross-platform computing technology needs to be designed. SUMMARY

[0005] To solve the existing technical problems, the present application provides a radar software cross-platform computing technology.

[0006] The specific content of the present application is as follows: a radar software cross-platform computing technology, which comprises abstracting radar information processing software components into vector calculation units, parallel calculation units, memory space allocation units, system function calling units, data type and size end conversion units, each unit is downwardly adapted to the heterogeneity of each platform, and upwardly provides a universal function calling interface.

[0007] Further, the vector calculation unit calling interface universalization in the radar information processing software comprises: uniformly describing the vector calculation unit, providing a unified function calling interface upwardly, and integrating each platform vector calculation library downwardly.

[0008] Further, the interface generalization of the parallel computing unit in the radar information processing software includes: generalizing the parallel computing unit, providing unified guidance instructions upward, and integrating the multi-core parallel framework of each platform downward.

[0009] Further, the interface generalization of the memory space allocation unit in the radar information processing software includes: generalizing the memory space allocation unit, providing unified function call interfaces upward, and integrating the memory operation library of each platform downward.

[0010] Further, the interface generalization of the system function call unit in the radar information processing software includes: packaging the related system call functions in the radar information processing software and automatically adapting to the corresponding platform to achieve flexible expansion.

[0011] Further, the interface generalization of the data type and size end unit in the radar information processing software includes: packaging the related data type definition and size end conversion in the radar information processing software and automatically adapting to the corresponding platform to achieve flexible expansion.

[0012] Further, the calling steps of the vector calculation unit include:

[0013] Step 1: Vector calculation library initialization, complete vector calculation library initialization operation through the initialization interface;

[0014] Step 2: Data block and view space binding, bind the data block address to the memory address, then bind the view on the data block address and get the data;

[0015] Step 3: Perform vector calculation operation, call related vector calculation functions;

[0016] Step 4: Destroy the view, then release the data block address space and destroy the data block;

[0017] Step 5: Vector calculation library end, complete the vector calculation library end operation through the end interface.

[0018] Further, the calling steps of the parallel computing unit include:

[0019] Step 1: Initialize the vector calculation library and the calculation slave task;

[0020] Step 2: Apply for the calculation slave task, apply for the number of calculation cores and core numbers required by the calculation task through the generalized interface;

[0021] Step 3: Call the related components of the radar information processing;

[0022] Step 4: Start the calculation slave task, release the calculation slave task after the calculation is completed;

[0023] Step 5: Exit the calculation from the task and end the vector calculation library.

[0024] Further, the memory space allocation unit, the system function calling unit, the data type and size end conversion unit adopt a platform dynamic adaptation mode, the upper layer provides a unified standard interface, the lower layer makes corresponding adaptation to the platform, and the heterogeneity of the platform is shielded, so that the development code and the platform are decoupled.

[0025] Further, the cross-platform calculation of the software comprises that: in the implementation code of different platforms, the interface provided by the vector calculation unit is called to replace the replaceable vector acceleration part; the interface provided by the parallel calculation unit is called to replace the replaceable parallel acceleration part in the vectorized code; the interface provided by the memory space allocation unit is called to replace the memory allocation management part in the memory space allocation unified code; the interface provided by the system function calling unit is called to replace the system function calling part in the system function calling unified code; the interface provided by the data type and size end conversion unit is called to replace the data type representation and size end conversion in the system function calling unified code; and cross-platform running is realized.

[0026] The radar information processing software cross-platform calculation technology of the application mainly comprises a vector calculation unit, a parallel calculation unit, a memory address allocation unit, a system function calling unit, a data type and size end conversion unit, a general interface is used to abstract the radar information processing software component unit, cross-platform transplantation of the radar information processing software is supported, and the radar information processing software is completed once the development function is transplanted. BRIEF DESCRIPTION OF DRAWINGS

[0027] The specific embodiments of the application will be further illustrated below in combination with the drawings.

[0028] Figure 1 The radar information processing software component unit framework is provided;

[0029] Figure 2 The vector calculation unit calling flow is provided;

[0030] Figure 3 The parallel calculation unit calling flow is provided;

[0031] Figure 4 The radar software cross-platform calculation technology application example is provided. DETAILED DESCRIPTION

[0032] As shown in the embodiment, the radar software cross-platform calculation technology is disclosed, which is based on the radar information processing software and comprises the following contents: Figure 1

[0033] ​(1) The radar information processing software is abstracted into a vector calculation unit, a parallel calculation unit, a memory address allocation unit, a system function calling unit, a data type and size end conversion unit.

[0034] The units of the radar information processing software are independent of each other, and provide interfaces of corresponding functions to the outside. According to the design idea of the middleware, each component unit is downwardly adapted to the heterogeneity of each platform, and upwardly provides a universal function calling interface, as follows:

[0035] (2) The vector calculation unit calling interface of the radar information processing software is universalized: the vector calculation unit is described in a universalized manner, a unified function calling interface is provided upwardly, and each platform vector calculation library is integrated downwardly.

[0036] (3) The parallel calculation unit interface of the radar information processing software is universalized: the parallel calculation unit is described in a universalized manner, a unified guidance instruction is provided upwardly, and each platform multi-core parallel framework is integrated downwardly.

[0037] (4) The memory space allocation unit interface of the radar information processing software is universalized: the memory space allocation unit is described in a universalized manner, a unified function calling interface is provided upwardly, and each platform memory operation library is integrated downwardly.

[0038] (5) The system function calling unit interface of the radar information processing software is universalized: the related system calling functions in the radar information processing software are packaged and automatically adapted in combination with the corresponding platform, so that flexible expansion can be realized.

[0039] (6) The data type and size end unit interface of the radar information processing software is universalized: the related data type definition and size end conversion in the radar information processing software are packaged and automatically adapted in combination with the corresponding platform, so that flexible expansion can be realized.

[0040] The calling flow of the vector calculation unit is shown in Figure 2 , and the steps are as follows:

[0041] First step: vector calculation library initialization, the vector calculation library initialization operation is completed through the initialization interface;

[0042] Second step: data block and view space binding, the data block address is bound to the memory address, and then the view binding is performed on the data block address and the data is acquired;

[0043] Third step: vector calculation operation, the related vector calculation function is called;

[0044] Fourth step: destroy the view, then release the data block address space and destroy the data block;

[0045] Fifth step: the vector calculation library ends, and the vector calculation library ending operation is completed through an ending interface.

[0046] The calling flow of the parallel computing unit is shown in Figure 3 The steps are as follows:

[0047] First step: initializing the vector calculation library and computing slave tasks;

[0048] Second step: applying for the computing slave task, and applying for the number of computing cores and core numbers required by the computing task through a generalized interface;

[0049] Third step: calling the radar information processing related components;

[0050] Fourth step: starting the computing slave task, and releasing the computing slave task after the computation is completed;

[0051] Fifth step: exiting the computing slave task and ending the vector calculation library.

[0052] The memory space allocation unit, the system function calling unit, the data type and size end conversion unit adopt a platform dynamic adaptation mode, the upper layer provides a unified standard interface, and the lower layer makes corresponding adaptation to the platform, so as to shield the heterogeneity of the platform and realize the decoupling of the development code and the platform.

[0053] The radar information processing software of the embodiment mainly includes a vector calculation unit, a parallel computing unit, a memory address allocation unit, a system function calling unit, a data type and size end conversion unit, the vector calculation unit abstracts the high-performance function library of the platform heterogeneity downwards; the parallel computing unit abstracts the code block of the platform heterogeneity downwards; the memory space allocation unit abstracts the memory allocation management algorithm of the platform heterogeneity downwards; the system function calling unit abstracts the operating system calling function of the platform heterogeneity downwards; the data type and size end conversion unit abstracts the data type representation and size end conversion implementation of the platform heterogeneity downwards; each unit above provides a generalized function interface upwards. The application adopts the design idea of middleware, each component unit shields the heterogeneity of the hardware platform downwards, and provides a unified development interface upwards, shortens the software development cycle, and realizes code reuse.

[0054] According to the radar information processing software, the processing process of an algorithm includes the following steps:

[0055] The vectorization part of the algorithm in the implementation code of multiple platforms is replaced by calling the interface provided by the vector calculation unit;

[0056] The parallel acceleration part of the vectorized code is replaced by calling the interface provided by the parallel computing unit;

[0057] The memory allocation management part in the parallelized code is replaced by an interface provided by the memory space allocation unit;

[0058] The system function calling part in the memory space allocation unified code is replaced by an interface provided by the system function calling unit;

[0059] The data type representation and endianness conversion in the system function calling unified code are replaced by an interface provided by the data type and endianness conversion unit;

[0060] After the above operations, the pulse compression source code can run across multiple platforms.

[0061] As shown in Figure 4 Pulse compression source code developed in combination with a pulse compression function algorithm is taken as an example, and is simulated and implemented on a Windows platform (64-bit), and runs on a blade platform (LINUX, 64-bit), a Huaru 2 (VXWorks6.8, 64-bit) and a T4240 platform (VXWorks7.0, 64-bit), and the components can all normally operate and the functions are correct.

[0062] The radar software cross-platform computing technology of the present application standardizes the abstract unit interface, reduces the development threshold, reduces the workload during code transplantation, and realizes cross-platform transplantation and operation of one function with one set of code. The present application reduces the complexity of application design by decoupling the developed software, and developers do not need to focus on the internal implementation of the vector calculation and parallel calculation units.

[0063] In the above description, many specific details are set forth in order to provide a thorough understanding of the present application. However, the above description is only a preferred embodiment of the present application, and the present application can be implemented in many other ways different from those described herein, and therefore the present application is not limited by the specific implementation disclosed above. Meanwhile, any person skilled in the art can make many possible changes and modifications to the technical solutions of the present application, or modify equivalent embodiments, by using the disclosed methods and technical contents, without departing from the scope of the technical solutions of the present application. Any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present application, without departing from the scope of the technical solutions of the present application, all still belong to the scope of protection of the technical solutions of the present application.

Claims

1. A radar software cross-platform computing system, characterized by: The radar information processing software is abstracted into a vector calculation unit, a parallel calculation unit, a memory space allocation unit, a system function calling unit, a data type and size end conversion unit, each unit is downwardly adapted to the heterogeneity of each platform and upwardly provides a universal function calling interface; The radar information processing software is abstracted into a vector calculation unit, a parallel calculation unit, a memory space allocation unit, a system function calling unit, a data type and size end conversion unit, each unit is downwardly adapted to the heterogeneity of each platform and upwardly provides a universal function calling interface; The radar information processing software is abstracted into a vector calculation unit, a parallel calculation unit, a memory space allocation unit, a system function calling unit, a data type and size end conversion unit, each unit is downwardly adapted to the heterogeneity of each platform and upwardly provides a universal function calling interface; The radar information processing software is abstracted into a vector calculation unit, a parallel calculation unit, a memory space allocation unit, a system function calling unit, a data type and size end conversion unit, each unit is downwardly adapted to the heterogeneity of each platform and upwardly provides a universal function calling interface; The radar information processing software is abstracted into a vector calculation unit, a parallel calculation unit, a memory space allocation unit, a system function calling unit, a data type and size end conversion unit, each unit is downwardly adapted to the heterogeneity of each platform and upwardly provides a universal function calling interface; The radar information processing software is abstracted into a vector calculation unit, a parallel calculation unit, a memory space allocation unit, a system function calling unit, a data type and size end conversion unit, each unit is downwardly adapted to the heterogeneity of each platform and upwardly provides a universal function calling interface; The vector calculation unit calling steps include: Step 1: Vector calculation library initialization, the vector calculation library initialization operation is completed through the initialization interface; Step 2: Data block and view space binding, the data block address is bound to the memory address, and then the view binding is performed on the data block address and the data is acquired; Step 3: Vector calculation operation, the related vector calculation function is called; Step 4: Destroy the view, then release the data block address space and destroy the data block; Step 5: Vector calculation library end, the vector calculation library end operation is completed through the end interface.

2. The radar software cross-platform computing system of claim 1, wherein: The parallel calculation unit calling steps include: Step 1: Initialize the vector calculation library and the calculation slave task; Step 2: Apply for the calculation slave task, the number of calculation cores and core numbers required by the calculation task are applied through the universal interface; Step 3: Call the radar information processing related component; Step 4: Start the calculation slave task, and release the calculation slave task after the calculation is completed; Step 5: Exit the calculation slave task and end the vector calculation library.

3. The radar software cross-platform computing system of claim 1, wherein: The memory space allocation unit, the system function calling unit, the data type and the size end conversion unit adopt the platform dynamic adaptation mode, the upper layer provides a unified standard interface, the lower layer adapts to the platform, the heterogeneity of the platform is shielded, and the development code is decoupled from the platform.

4. The radar software cross-platform computing system of claim 1, wherein: The cross-platform computing of the software comprises: replacing an interface provided by a vector computing unit in a vector-accelerable part of implementation code of different platforms; replacing an interface provided by a parallel computing unit in a parallel-accelerable part of the vectorized code; replacing an interface provided by a memory space allocation unit in a memory allocation management part of the parallelized code; replacing a system function calling part in the memory space allocation unified code into an interface provided by a system function calling unit; replacing a data type representation and endianness conversion part in the system function calling unified code into an interface provided by a data type, endianness conversion unit; and realizing cross-platform running.

Images