Statistical delay QoS driven resource allocation and performance analysis for wireless communication networks

Delay quality-of-service (QoS) guarantees play a critical role in enabling delay-sensitive wireless applications. By applying the theory of effective capacity (EC), the maximum arrival rate with a guaranteed delay-outage probability constraint, is analyzed and investigated in terms of delayconstrained resource allocation and link-layer throughput analysis. Firstly, a joint optimization problem of link-layer energy efficiency (EE) and EC in a single-user single-carrier communication system, is proposed and investigated, under a delay violation probability requirement and an average transmit power constraint. Formulated as a normalized multiobjective optimization problem (MOP), the problem is transformed into a weighted single-objective optimization problem (SOP), and then solved. The proposed optimal power value is proved to be sufficient for the Pareto optimal set of the original EE-EC MOP. Secondly, a total EC maximization problem subject to the individual linklayer EE requirement as well as the per-user average transmit power limit, in a multi-user multi-carrier orthogonal frequency-division multiple access (OFDMA) system, is proposed and analyzed. Formulated as a combinatorial integer programming problem, the problem is decoupled into a frequency provisioning problem and an independent per-user multi-carrier EE-EC tradeoff problem. A low-complexity heuristic algorithm is proposed to obtain the subcarrier assignment solution coupled with a per-user optimal power allocation strategy, across frequency and time domains. Finally, the achievable link-layer rate under the per-user delay QoS requirements is studied for a downlink M-user non-orthogonal multiple access (NOMA) network. The impact of the transmit signal-to-noise ratio (SNR) and the delay QoS requirement on the per-user achievable EC and the total link-layer rate is investigated and compared between NOMA and orthogonal multiple access (OMA) networks. All theoretical conclusions and closed-form expressions are confirmed with Monte Carlo results