SNC-Enabled Delay Analysis and DRL-Based Power Control for RSMA Systems

Rate-splitting multiple access (RSMA) technology is expected to play an important role in next-generation wireless networks; however, the delay analysis of RSMA systems has rarely been addressed. In this paper, a downlink RSMA system is studied and its performance is evaluated by Mellin transform analysis. Using stochastic network calculus (SNC), we derive closed-form expressions for the delay violation probability (DVP) under various fading conditions, including Rayleigh, Nakagami-<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$m$</tex-math></inline-formula>, and Ricean channels. These expressions are proven to converge and provide theoretical upper bounds for the DVP. Simulation results fully verify the correctness of these theoretical formulas through Monte Carlo simulation. To characterize performance under scenarios of strict delay sensitivity and limited energy resources, we define the system utility (SU) as the weighted sum of energy efficiency (EE) and the negative of DVP. We then formulate a joint optimization problem aimed at maximizing SU and propose a power allocation strategy to satisfy the delay requirements and energy efficiency indicators of all users. To cleverly adapt to dynamic channel conditions, we introduce a near-optimal power allocation scheme based on deep deterministic policy gradient (DDPG) to approximate the capacity of the network. In addition, the simulations illustrate that the proposed scheme increases EE by about 25% and reduces the DVP by nearly 10 orders of magnitude at a 1 ms delay threshold compared with non-orthogonal multiple access (NOMA) system, especially as the number of users increases.