A fixed dataflow sub-lattice based parallel reduction algorithm for MIMO detection

A low complexity fixed dataflow lattice reduction algorithm is proposed for MIMO detection in this paper. We first speed up the well-known Lenstra Lenstra Lovász (LLL) algorithm with an additional optimal swap selection (OSS) step, which helps to achieve LLL-reduced basis with much fewer LLL loops. The problem with the OSS-LLL algorithm, so as the original LLL algorithm, is that it suffers from variant dataflow and complexity, which are two key impairments for hardware implementation of real time communication systems. The sub-lattice based odd-even-alternate reduction method is further employed to fix the dataflow and the complexity, at the same time it can be easily implemented in parallel on single instruction multiple data (SIMD) platform. Even for non-parallel implementation, although the convergence speed of the proposed algorithm is slower compared to the OSS-LLL algorithm, the additional OSS operations during the LR process can be saved which results in a relatively lower complexity. Simulation results verify our analysis and show that the proposed algorithm achieves almost the same performance as the LLL aided methods for MIMO detection.