System and method for real-time electromagnetic spectrum analysis driven by gpu data stream

Abstract

The invention relates to a GPU-based data stream-driven real-time electromagnetic spectrum analysis system and method. A high-performance GPU is adopted as a computing and processing core of a real-time spectrum measurement and analysis device or system; a high-speed display memory inside the high-performance GPU is used as a cache of the device or the system; a high-speed bus inside the GPU replaces the bus of the device or the system; a data transmission link is set inside the GPU; a data stream driving mode is adopted to realize high-speed transmission, time domain-frequency domain conversion, digital filtering, digital signal conditioning, data compression, display processing and other functions of digital signals after sampling, and thus, synchronization requirements of signal processing and test sampling in real-time spectrum measurement and analysis are met. The analysis system disclosed in the invention can greatly improve the analysis and processing capability of digital signals so as to match the signal sampling speed, buffer overflow caused by untimely processing and signal loss can be avoided, and real-time spectrum measurement and analysis on the broadband or transientelectromagnetic signals can be further realized.

Description

technical field [0001] The invention relates to a real-time electromagnetic spectrum analysis system and method driven by a GPU data flow, and belongs to the field of electronic signal test and measurement. Background technique [0002] In order to achieve greater capacity and better security, modern electronic information systems occupy more and more bandwidth and the rate of signal change is faster and faster, which makes it necessary to measure and analyze the characteristic changes of the spectrum over a wide frequency band technical means. Real-time spectrum analyzer is an important tool to realize this method. On the other hand, in remote control telemetry and remote sensing, it is often necessary to capture and analyze the spectrum characteristics of burst signals. The shorter the duration of the burst signal and the faster the change, the wider the instantaneous bandwidth will be occupied. Only the real-time spectrum analyzer has the ability to analyze the burst sig...

AI Technical Summary

Problems solved by technology

Control flow drive is used to realize the data interaction of each part. All data interaction requires the CPU to control the memory to be read through the computer bus, but the computer bus contains not only the interactive data, but also the control instruction set sent by the controller to other parts. The bus bandwidth actually used for data transmission is only half of the total bandwidth. When a larger amount of sampled data needs to be processed, this architecture cannot meet the requirements of real-time test analysis

Although GPU can solve the problem of insufficient computing resources of FPGA or DSP devices, it still needs CPU to control memory to transfer and transmit data during data processing, which cannot get rid of the defects of traditional architecture

The traditional architecture has wide versatility, but it is not the optimal architecture for the single high-speed data processing requirement of real-time spectrum analysis

Images