Experimental SEFDM Pipelined Iterative Detection Architecture with Improved Throughput

Abstract

In spectrally efficient frequency division multiplexing (SEFDM), the separation between subcarriers is reduced below the Nyquist criteria, enhancing bandwidth utilisation in comparison to orthogonal frequency division multiplexing (OFDM). This leads to self-induced inter-carrier interference (ICI) in the SEFDM signal, which requires more sophisticated detectors to retrieve the transmitted data. In previous work, iterative detectors (IDs) have been used to recover the SEFDM signal after processing a certain number of iterations, however, the sequential iterative process increases the processing time with the number of iterations, leading to throughput reduction. In this work, ID pipelining is designed and implemented in software defined radio (SDR) to reduce the overall system detection latency and improve the throughput. Thus, symbols are allocated into parallel IDs that have no waiting time as they are received. Our experimental findings show that throughput will improve linearly with the number of the paralleled ID elements, however, hardware complexity also increases linearly with the number of ID elements.

The First 15 Years of SEFDM: A Brief Survey

“…2016 [95] 2.4 Gbit/s WDM visible light communication (VLC) experiment Sep. 2016 [64] VLC experiment using RLS time-domain channel estimation May 2017 [23] Transmission experiment for carrier aggregation in a wireless channel Oct. 2017 [96] Over-the-Air testing using software defined radio USRP RIO devices Jan. 2018 [97] Experimental implementations of real-time systems in fading channel Jun. 2018 [98] Pipelined iterative detection experiment Jun. 2018 [99] Self interference cancellation experiment Jul.…”

EVM Loss: A Loss Function for Training Neural Networks in Communication Systems

“…Of the latter, of the most promising is spectrally efficient frequency division multiplexing (SEFDM), a modification of orthogonal frequency division multiplexing (OFDM) first proposed in 2006 by Rodrigues and Darwazeh in [ 4 ]. In SEFDM, orthogonality between subcarriers is broken by compressing the overall bandwidth at the cost of inter-carrier interference (ICI) [ 5 ], and machine learning is a promising topic to compensate for this.…”

Neural Network Equalisation and Symbol Detection for 802.11p V2V Communication at 5.9GHz

“…Introducing such a compression improves spectral usage, which is advantageous, but causes self-induced deterministic inter-carrier interference (ICI). As a result of this, over the years, research has been conducted that has resulted in a compression of 30% at reasonable bit-error rates (BERs) using 4-level quadrature amplitude modulation (4-QAM) [13]. The result of the ICI means that complex receivers are required to recover the symbols at the receiver, and to date sphere decoders (SDs) are the most popular choice to take advantage of the newly improved spectral efficiency.…”