Combining Run-Length Encoding Preprocessing With Lempel-Ziv-Markov Algorithm to Enhance Compression for Automotive Radar Data

The integration of high-resolution radars in vehicles with central processing systems has significantly increased data volume in sensor networks. To address this issue, researchers have investigated radar data compression methods designed to reduce information loss, conserve resources, and improve processing efficiency. These methods include lossy preprocessing and quantization. This paper focuses on a data stream generated by Run-Length Encoding (RLE), which is adapted for complex radar data with a dynamic counter length. Following this, the data stream is further compressed using either the Lempel-Ziv-Markov Algorithm (LZMA) directly or a combination of Huffman coding before LZMA. We compare the compression ratio achieved by RLE alone, RLE with Huffman coding, RLE with LZMA, and RLE with Huffman coding followed by LZMA. Our methods achieve average compression ratios of up to 48500 on actual radar data.