Semi-Rate-Splitting Multiple Access (SRMA)

Abstract

In single-input single-output downlink systems, the common and private streams in rate-splitting multiple access (RSMA) fully overlap in the power domain, causing strong inter-private interference. Under imperfect successive interference cancellation (SIC), this interference significantly increases users’ decoding error probability. To overcome this limitation, we propose a semi-rate-splitting multiple access (SRMA) scheme that distributes private streams across two distinct bandwidth/time regions rather than fully overlapping them. To evaluate the proposed SRMA, we formulate two design problems: 1) maximizing the minimum ergodic capacity (EC) among users and 2) minimizing the system connection outage probability (COP). Both problems jointly optimize the rate-split ratio, power allocation coefficient, and partitioned-resource factor. To address the non-convexity in these problems, we employ successive convex approximation and analytical reformulation to relax them into tractable convex problems with closed-form solutions. Monte-Carlo simulations demonstrate that SRMA outstandingly improves COP performance while achieving EC levels comparable to RSMA and outperforming conventional non-orthogonal multiple.