Traffic Aware Dynamic Resource Management And Optimization In Wireless Networks

With the rapid development of smart mobile equipment, such as smart phones, more and more people are getting data service through wireless mobile networks. In this situation wireless networks is facing with not only serving booming data traffic but also the requirement of improving quality of service e.g. lowering data transfer delay. To deal with these requirements, on the one hand advanced transmission technologies such as Orthogonal Frequency Division Multiplexing (OFDM) are applied to increase the physical layer bit rate; on the other hand, frequency reuse one, dense deployment and heterogeneous network are used to enhance the capacity and lower the delay of data traffic transfer. Note that the large number of deployed base stations (BS) causes the increasing of energy consumption. So reducing the energy consumption is also important. In practical wireless network, the data traffic is spatial and temporal non-uniform distributed. The non-uniform spatial distribution of data traffic may cause load unbalance across the network and the increasing of data transfer delay. On the other side, the non-uniform temporal distribution of data traffic may cause energy wastage during the period when little traffics are served. So we consider the traffic aware dynamic resource management of wireless network to achieve the goal of network performance improvement and energy saving.First, this thesis focuses on the spatial non-uniform distribution of data traffic and proposes a traffic aware dynamic sub-band allocation algorithm to make spectrum resource adapt to the spatial distribution of traffic. Through allocating sub-band to each BS across the cellular networks, we optimize the average data transfer delay. We use flow-level model to capture the dynamics of wireless data transfer and get the stochastic performance such as delay, load. Then based on fractional frequency reuse for inter-cell interference coordination, proposed algorithm gets the sub-band allocation through the method of graph max weighted independent set coloring. The proposed algorithm can balance the network load and coordinate interference at the same time. Compared with existing works dealing spatial non-uniform traffic using dynamic user association, this algorithm can avoid users served by neighbor BS which leads to low signal strength and high interference especially. Analysis and simulations show this algorithm can lower the data transfer delay especially for cell edge users. Then we focus on the temporal non-uniform distribution of data traffic. In heterogeneous networks we propose a traffic aware small cell sleeping scheme to make energy resource adapt to the temporal distribution of data traffic. The scheme utilizes the interval when no data traffic arrives and puts the BS into sleep to achieve energy saving. In this scheme micro BS will dynamically switch the state between active and sleep according to the ongoing flow number and the sleep control parameter. The scheme designs a macro BS’s cooperation to enhance energy saving. We use flow-level model to capture the dynamic process of micro BS serving data traffic and sleeping. Then we get the energy and delay performance through a double layer Markov state transfer diagram and solving global balance equations. Finally using the optimal energy-delay tradeoff curve, we get the optimal sleep control parameter which can get the maximal energy saving while meeting the requirement for average delay. Analysis and simulations show this scheme can get energy consumption reduced.