Frame-Level Cross-Layer Power Optimization for Uplink Wireless Low-Latency Streaming

This paper proposes a frame-level cross-layer power optimization for uplink wireless low-latency video streaming. The study begins by analyzing the feasibility of Frame-Level Cross-Layer Optimization (FCO). The first- and second-order moments of the Signal-to-Interference-plus-Noise Ratio under Maximum Ratio (MR), Zero-Forcing (ZF), and Minimum Mean Squared Error (MMSE) are derived. Based on this, we further analyze the probability that the average transmission rate deviates from its expectation. The results show that ZF and MMSE, even with a few antennas, ensure the precise characterization of the average rate using its expected value at the frame-level time scale, making FCO feasible. Building on this analysis, we propose the Frame-Level Cross-Layer Mechanism (FCM) that reserves a few frames in the sender’s buffer to enable precise power control while addressing the challenges of QoS estimation inaccuracy. A non-convex cross-layer optimization for joint power control is then formulated, leveraging application-layer frame size and physical-layer average rate. We convert non-convex terms to convex ones, with the resulting non-convex term approximated by Taylor expansion. The Successive Convex Approximation is used to solve the converted problem. Numerical results show that the proposed FCM and joint power control reduce transmission power while meeting the low-latency requirements for live streaming.