A Rate-Splitting Strategy to Enable Joint Radar Sensing and Communication with Partial CSIT

Rate-Splitting Multiple Access: Fundamentals, Survey, and Future Research Trends

“…RSMA not only better manages the interference among communication users, but also successfully handles the interference between communication and radar. The authors of [140] further show the tradeoff gain of RSMA-aided DFRC for imperfect CSIT. The authors of [139] further consider low resolution digital-to-analog converter (DAC) units for RSMAaided DFRC.…”

“…In [84], a globally optimal algorithm, namely, the successive incumbent transcending (SIT) branch and bound (BB) algorithm was proposed to solve P 1 optimally, when the WSR, EE, and sum-power objective functions were considered. To reduce the computational complexity, a large number of works on RSMA have focused on developing suboptimal algorithms to solve P 1 , P 2 , and P 3 with affordable and tractable complexities, such as weighted minimum mean square error (WMMSE)-based algorithms [53], [58], [77], [83], [87], [90], [99], [106], [109], [119], [128], [137], successive convex approximation (SCA)based algorithms [74], [82], [89], [96], [100], [102], [104], [107], [108], [114]- [116], [122], [123], [126], [127], [135], alternating direction method of multipliers (ADMM)-based algorithms [117], [139], [140], and semidefinite relaxation (SDR)-based algorithms [73], [86], [105], [124], [125]. These algorithms are able to converge to the Karush-Kuhn-Tucker (KKT) points of the original problem.…”

“…• Intelligent reconfigurable surface (IRS) [126], [127], [134]. • Joint communication and radar [117], [139], [140].…”

“…2) Imperfect CSIT: Precoder optimization for RSMA has also been studied extensively for imperfect CSIT with the objective of maximizing the WSR [58], [77], [82], [83], [86], [90], [107], [109], [128], [135], [140], maximizing the minimum rate among users [56], [67], [87], [125], [137], maximizing the EE [131], and minimizing the sum-power consumption [73], [105]. For imperfect CSIT, the BS only has the knowledge of channel estimate h k for each instantaneous CSI vector h k of user-k. A typical imperfect CSIT model is given as follows…”

“…Long-term precoder optimization: When the CSIT estimation error h k is unbounded, for example, when the transmitter only knows the Gaussian distribution of the CSIT estimation error, a worst-case precoder optimization is not possible any more. Instead of designing precoders for each instantaneous channel, in this case, long-term RSMA precoder optimization over a long sequence of fading states is more commonly used, see, e.g., [58], [77], [83], [87], [90], [107], [109], [128], [135], [137], [140]. By defining H = [h 1 , .…”