BER-Driven Resource Allocation for Scalable Bitstreams Over OFDMA Networks

Conventional resource-allocation schemes in orthogonal frequency-division multiple-access (OFDMA) systems consider only one bit-error-rate (BER) requirement. When these schemes are applied to transmit bitstreams with multiple BER requirements, low resource utilization will occur since the target BER has to be the strictest one. We develop a BER-driven resource allocation for generic bitstreams (BRA-G) algorithm by exploiting multiple predetermined target BERs. Then, this scheme is extended to transmission of realistic scalable bitstreams (BRA-SB) where more important (MI) layers possess lower target BERs as compared with less important (LI) layers. The objective here is to maximize the expected quality of all JPEG 2000 (J2K) scalable bitstreams subject to a total power constraint. Instead of using static target BERs for different layers, BRA-SB water-fills over all layers to determine the number of transmit layers and the target BER of each layer simultaneously based on the importance of each layer in terms of reconstructed quality and average channel conditions. Bit-level results show that BRA-SB provides significant image quality improvement over a scheme that ignores the layer´s importance in terms of target BER and a BRA-SB-like scheme with static target BERs. A suboptimal algorithm is also proposed to reduce the computational complexity and its performance is close to the optimal one.