Distributing Complexity: A New Approach to Antenna Selection for Distributed Massive MIMO

Abstract

Antenna selection in Massive MIMO (Multiple Input Multiple Output) communication systems enables reduction of complexity, cost and power while keeping the channel capacity high and retaining the diversity, interference reduction, spatial multiplexity and array gains of Massive MIMO. We investigate the possibility of decentralised antenna selection both to parallelise the optimisation process and put the environment awareness to use. Results of experiments with two different power control rules and varying number of users show that a simple and computationally inexpensive algorithm can be used in real time. The algorithm we propose draws its foundations from selforganisation, environment awareness and randomness.

Distributed Antenna Selection for Massive MIMO Using Reversing Petri Nets

“…The algorithm was tested in the same conditions as the NN algorithm in [6], using the raytracing Matlab tool Ilmprop [12] with 64 distributed transmitters ( Fig. 2(a)) in the area which included 75 scatterers and one large obstacle.…”

“…The need for a strategic antenna selection in all observed scenarios is demonstrated by the results of significantly underperforming random selection. In [6], its has been shown that the distributed algorithms such as NN are resistant to errors in CSI and that they perform well even with just a (randomly selected) subset of subcarriers used for optimisation. We performed the same test for the RPN solution, and the result is shown in Fig.…”

“…In NN, the number of antennas for which the system reaches the best performance is an emergent property and as such cannot be controlled. Furthermore, in [6] it was seen that the NN algorithm requires more than N R antennas to serve N R users; RPN gives meaningful results and good performance even at N R antennas selected.…”

“…The optimisation problem in (1) has two variables, the subset of selected antennas and the optimal power distribution over them. Following the practice from [9], [6], we start from the assumption of P having all diagonal elements equal to 1/N R (their sum is unity, making the total power equal to ρN R /N T S ), and after the antenna selection optimise P by water filling for zero forcing. This was suggested in [9] as a practicality measure; hence the results in this paper are presented after linear beamforming, and with coherent data transmission in mind.…”

“…In a different perspective, random antenna selection was proposed as a computationally inexpensive alternative with satisfying results in some scenarios [5]. Finally, distributed decision-making has been suggested as a natural approach to distributed massive MIMO antenna selection [6], ensuring that highly correlated spatially clustered antennas are not prioritised in selection.…”