Abstract
Wireless networks beyond 5G will mostly be serving myriads of sensors and other machine-type communications (MTC), with each device having different requirements in respect to latency, error rate, energy consumption, spectral efficiency or other specifications. Multiple-input multiple-output (MIMO) systems remain a central technology towards 6G, and in cases where massive antenna arrays or cell-free networks are not possible to deploy and only moderately large antenna arrays are allowed, the detection problem at the base-station cannot rely on zero-forcing or matched filters and more complex detection schemes have to be used. The main challenge is to find low complexity, hardware feasible methods that are able to attain near optimal performance. Randomized algorithms based on Gibbs sampling (GS) were proven to perform very close to the optimal detection, even for moderately large antenna arrays, while yielding an acceptable number of operations. However, their performance is highly dependent on the chosen “temperature” parameter (TP). In this paper, we propose and study an optimized variant of the GS method, denoted by triple mixed GS, and where three distinct values for the TP are considered. The method exhibits faster convergence rates than the existing ones in the literature, hence requiring fewer iterations to achieve a target bit error rate. The proposed detector is suitable for symmetric large MIMO systems, however the proposed fixed complexity detector is highly suitable to spectrally efficient adaptively modulated MIMO (AM-MIMO) systems where different types of devices upload information at different bit rates or have different requirements regarding spectral efficiency. The proposed receiver is shown to attain quasi-optimal performance in both scenarios.
Highlights
Wireless networks are heading towards their 6th generation (6G), shifting from the scenario where most users were mobile phones, to one where a paraphernalia of sensors and other devices will lead to pervasive machine-type communications (MTC), which will constitute the dominant type of wireless links [1,2,3]
Most proposals for the generation rely on the existence of a massive Multiple-input multipleoutput (MIMO) base-station (BS) [6] (today widely considered to be systems with larger than 64 antenna elements [7], (p. 155)) or cell-free networks [8], and they rely on asymmetric MIMO cases, where the number of terminals is much lower than the number of antenna elements at the BS
Millimeter-waves systems are on the rise, but those tend to use low-order modulation due to both the high attenuation channels and the challenges posed to the analogue-to-digital converters (ADCs) when using wideband signaling
Summary
Wireless networks are heading towards their 6th generation (6G), shifting from the scenario where most users were mobile phones, to one where a paraphernalia of sensors and other devices will lead to pervasive machine-type communications (MTC), which will constitute the dominant type of wireless links [1,2,3]. Finding efficient detection algorithms for such systems has been the focus of some literature; for example, in [20], the authors proposed SD-based methods for optimal decoding of AM-MIMO, even though it is not a scalable solution for large MIMO due to the rising complexity that SD entails Advanced statistical methods, such as Markov chain Monte Carlo (MCMC), have already shown significant gains in signal processing for wireless communications. Numerical results confirm that the proposed variant outperforms both aforementioned works, and that by using a multiple restarts (MR) approach, quasi-optimal performance can be obtained without further increasing processing time Following this introductory section, the paper proceeds with the definition of the MIMO channel model, which makes use of the mapping of the MIMO communication problem in the complex-domain to one in the real-domain, which can be tackled by the GS method.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
