Design Exploration & Enhancements for Low Complexity Massive MIMO Detectors with High Modulation Order

Global energy consumed by communication and information technologies is expected to increase rapidly due to continuous usage of wireless standards and the expansion for their requirements [1]. In the next generation wireless communications, Multi Input and Multi Output (MIMO) systems are most promising technology to achieve high spectral efficiencies, while going past various challenges like resource and energy constraints [2]. There exists many detection algorithms like Maximum Likelihood (ML), Zero Forcing (ZF), Minimum Mean Square Error (MMSE) which have low silicon complexity but consume significant power for high-end MIMO systems, due to their high computational complexity. And then there are certain low power detection algorithms like real domain breadth first search K-best, with either conventional enumeration or Schnorr Euchner (SE) based enumeration. This improvement through either, comes with cost of comparatively high silicon complexity and sacrifices the performance in terms of detection bit error rate (BER). The complex domain equivalent may improve the BER performance but it’s dedicated algorithm ensures even higher silicon complexity. Several modifications have been performed on original complex domain K-best algorithm to decrease its high silicon complexity, retaining the better performance of the system. This work focuses on study and implementation of original real SE based K-best algorithm [3]. It also features my attempt to perform theoretical analysis of original complex domain detection algorithm, and to implement modified [4] and improved versions of complex domain to decrease its high silicon complexity, retaining BER performance. This work also focuses on exploration and implementation of past attempts on design modifications of complex domain algorithms and compare them across different attributes such as performance, computational and silicon complexity. Few system level and algorithmic level enhancements have been proposed and implemented for low complexity detectors explored. Dynamic fixed point iterative version of original real domain detector [3] has been studied and implemented, along with possible enhancements for complex domain detector. Pipelined hardware architecture of real domain SE based K-best detector [5] has also been studied as part of this work, with the intention of extending this to dynamic fixed point version and also complex domain detector