Date Added: Aug 2009
It is well-known that cross-layer scheduling which adapts power, rate and user allocation can achieve significant gain on system capacity. However, conventional cross-layer designs all require Channel State Information at the base station (CSIT) which is difficult to obtain in practice. In this paper, the authors focus on cross-layer resource optimization based on ACK/NAK feedback flows in OFDM systems without explicit CSIT. While the problem can be modeled as Markov Decision Process (MDP), brute force approach by policy iteration or value iteration cannot lead to any viable solution. Thus, they derive a simple closed-form solution for the MDP cross-layer problem, which is asymptotically optimal for sufficiently small target Packet Error Rate (PER).