Distributed Antenna Selection for Massive MIMO Using Reversing Petri Nets

Abstract

Distributed antenna selection for Distributed Massive MIMO (Multiple Input Multiple Output) communication systems reduces computational complexity compared to centralised approaches, and provides high fault tolerance while retaining diversity and spatial multiplexity. We propose a novel distributed algorithm for antenna selection and show its advantage over existing centralised and distributed solutions. The proposed algorithm is shown to perform well with imperfect channel state information, and to execute a small number of simple computational operations per node, converging fast to a steady state. We base it on Reversing Petri Nets, a variant of Petri nets inspired by reversible computation, capable of both forward and backward execution while obeying conservation laws.

Software and Reversible Systems: A Survey of Recent Activities

“…In the first example [64], we perform antenna selection in a large distributed antenna array which serves as a distributed base station in a next generation cellular network: at any point in time, we want to use n out of m available antennas to serve k < n users in the cell. The subset of antennas to be used is selected so to maximise the Shannon capacity of the communication channel between the base station and the users, which is a non-trivial optimisation task: selecting simply the antennas with the strongest signal does not help as they tend to be correlated and not contributing to the diversity in the channel.…”

Reversible Computation in Wireless Communications

“…As this chapter focuses on applications, the reader interested in details about reversing Petri nets used in this example is advised to see [18]. The presentation here is based on [22].…”

Controlling Reversibility in Reversing Petri Nets with Application to Wireless Communications

“…Results of the RPN-based approach on an array consisting of 64 antennas serving 16 users, varying the number of selected antennas from 16 to 64 are shown in Fig. 2 [16]. If we run five RPN models in parallel and select the one with the best performance for the final selection, the results are consistently superior to those of a centralised (greedy) algorithm, and if we run just one (equivalent to the average of the performance of these five models) the results are on par with those of the centralised algorithm.…”