Abstract
A joint channel adaptive rate control and randomized scheduling algorithm based on learning automata (LA) K. S. Narendra et al., (1989)is presented. The scheduling is performed at the medium access control (MAC) layer whereas the rate selection takes place at the physical/link (PHY/LINK) layer. The two components residing in the two layers exchange minimal amount of information and adaptively achieve the best throughput and desired quality of service (QoS) in terms of average transmission rates in the prevailing channel conditions. Scheduling is carried out by a LA of continuous reward penalty variate, and a discrete pursuit reward inaction (DPRI) type B. J. Oommen et al., (1990) is used for adaptive rate selection. While simple to implement, this technique requires no explicit channel estimation phase. The only feedback required are the single bit ACK signal indicating the correct reception of packets. As shown in the convergence theorems, the algorithm achieves optimal performance in "stationary'" channels. With slowly varying channels, the MCS selection algorithm sees a "quasistationary" channel and adaptively converges to the optimality. Simulation results are provided for parameters as per to HSDPA standard.