Precoding Design for Multi-User MIMO Joint Communications and Sensing

TL;DR

This work investigates precoding design for MU-MIMO joint communications and sensing (JCAS) under potential interference between sensing and communication channels. It derives performance indicators for sensing and communications, notably a closed-form detector statistic for target detection via an NP CFAR detector and SINR expressions for both separate and shared signaling, highlighting how the transmit signal's kurtosis constrains sensing. The study shows that reusing communication signals for sensing can mitigate interference effects on communications, but sensing performance is limited by the kurtosis of the transmit constellation, with Gaussian signals offering some invariance to beamforming choices. An autoencoder-based simulation framework is used to explore precoding and signaling trade-offs, providing practical guidance: use a constant-modulus signal for sensing when interference is absent, and, if interference is present, leverage the communication signal for sensing while acknowledging the kurtosis trade-off; avoid combining multiple sensing signals as it degrades detection. The results underscore the kurtosis of transmit signals as a central design parameter in JCAS and offer actionable guidance for transmit design in MU-MIMO JCAS systems.

Abstract

We investigate precoding for multi-user (MU) multiple-input multiple-output (MIMO) joint communications and sensing (JCAS) systems, taking into account the potential interference between sensing and communication channels. We derive indicators for the sensing and communication performance, i.e., the detection probability and the communication signal-to-interference-and-noise ratio (SINR) for general input signals. Our results show that the use of the communication signal for sensing can prevent a loss in communication performance if channel interference occurs, while the kurtosis of the transmit alphabet of the communication signal limits the sensing performance. We present simulation results of example setups.

Figures (7)

• Figure 1: JCAS system. The sensing target can become a reflector in the communication channel of UE1, introducing interference $h_{\text{target}}$.
• Figure 2: Communication SINR with $\text{SNR}_{\text{c}} =20\,$dB and $h_1=1$.
• Figure 3: Statistics for detection with $\sigma_{\text{s}}^2/\sigma_{\text{ns}}^2=-5\,$dB and a 16-QAM signal. The dashed line indicates the detection threshold for $P_{\text{f}}=0.01$.
• Figure 4: Detection probability with $\sigma_{\text{s}}^2/\sigma_{\text{ns}}^2=-5\,$dB for phase-modulated and 16-QAM sensing input. The dashed line indicates the operating point of $P_{\text{f}}=0.01$ used for further evaluation.
• Figure 5: Beam patterns for two UE using 16-QAM with example scenarios \ref{['plot:b0']}$\tilde{\mathsf{z}}_{\mathrm{s},\text{CM}}$, \ref{['plot:b12']}$\tilde{\mathsf{z}}_{\mathrm{s},\text{QAM}+\text{CM}}$, and \ref{['plot:b1']}$\tilde{\mathsf{z}}_{\mathrm{s},\text{QAM}}$.