An investigation in broadband network : admission control, power allocation, subcarrier assignment

Abstract

Broadband wireless network providing multiple services is investigated in this thesis. The quality of service is provided by a developed prediction algorithm employing Bayesian Neural Network. The algorithm referred as QoS predictor is evaluated based on a formerly developed Tri-Threshold Bandwidth Reservation Call Admission Control Scheme. The predictor helps to improve the resource management of high speed packet networks, thus saving the cost and bandwidth. When considering about the transmission level, orthogonal carrier combined with multiple access is a scheme enabling high–speed data, video, and multimedia communication and is used by a variety of commercial broadband systems. Moreover, OFDMA facilitates relaying in multiple hop networks, which provides cooperative diversity. Algorithms of subcarrier allocation and power allocation based on multiuser OFDMA relaying network are proposed in this thesis. Numerical results are conducted to show the merits in terms of throughput achievement of the proposed algorithm—optimal-BER and water-filling over uniform power allocation. It is verified that by employing optimal-BER, not only improved aggregate throughput is achieved, but also the BER requirements can be satisfied under adaptive modulations.