Investigate Redundancy In Sounding Reference Signal Based Channel Estimates

Sammanfattning: 5G supports enormous increase in data rate. Massive antenna beamforming is expected to play a key role in increasing capacity in case of multi-user MIMO and coverage in case of single-user MIMO. The large number of antennas in massive MIMO system will lead to enormous amount of channel state information being stored in the memory and this necessitates the use of compression techniques for efficient utilization of memory, which is limited. Sounding Reference Signals (SRS) are transmitted in the uplink to obtain channel estimate. In TDD based systems, by exploiting channel reciprocity channel estimates received in the uplink can be used in downlink as well. The product, we work on at Ericsson, is a TDD based system and uses SRS based channel estimates to compute beamforming weights to facilitate massive antenna beamforming. SRS based channel state information is represented by 32-bit complex number in this system, which is received per Evolved Node B (eNodeB) antenna, per User Equipment (UE) transmission antenna, and per Physical Resource Block Group (PRBG). This results in a significant amount of data that needs to be stored in the eNodeB. However, memory in the Digital Unit of eNodeB is limited. SRS based estimates occupy a major portion of this memory and therefore limit the capacity of the eNodeB for beamforming. This thesis focuses on the evaluation and implementation of lossless and lossy compression of SRS based channel estimates to attain space savings in the shared memory of eNodeB. This will help in achieving higher capacity for reciprocity-based beamforming and prolong the lifetime of existing hardware. Performance of various lossless data compression algorithms was analyzed based on compression ratio, speed and complexity and the optimal one was selected. Lossy compression of SRS based channel estimates was also implemented for LOS UEs using linear regression by least squares estimate. Impact on performance due to application of lossy compression algorithm was studied.