In wireless networks with centralized base stations, cross-layer scheduling which adapts user selection, power allocation and rate allocation, can achieve a tremendous gain. However, conventional cross-layer designs all require channel state information at the base station (CSIT) which is difficult to obtain in practice. Cross-layer designs requiring no CSIT are mostly modelled as a Markov Decision Process (MDP). Yet, complicated searching algorithms were involved and no joint optimization of power, rate and user selection have been considered. In this thesis, we formulated the goodput maximization problem in optimization framework and provided solutions and analysis for 2 special cases and one general case which is formulated as a MDP. We proposed a cross-layer scheduler which gives a...[ Read more ]

In wireless networks with centralized base stations, cross-layer scheduling which adapts user selection, power allocation and rate allocation, can achieve a tremendous gain. However, conventional cross-layer designs all require channel state information at the base station (CSIT) which is difficult to obtain in practice. Cross-layer designs requiring no CSIT are mostly modelled as a Markov Decision Process (MDP). Yet, complicated searching algorithms were involved and no joint optimization of power, rate and user selection have been considered. In this thesis, we formulated the goodput maximization problem in optimization framework and provided solutions and analysis for 2 special cases and one general case which is formulated as a MDP. We proposed a cross-layer scheduler which gives a closed-form solution and provides QoS guarantee by optimal user selection, power and rate allocation. This algorithm achieves 89% of the capacity of the frequency selective channel.