Available Bandwidth Estimation And Prediction In Multi-hop Wireless Networks

For the purpose of ubiquitous networking, the multi-hop wireless network has been a vital extension to wired networks. With the widespread use of multimedia applications that require Quality of Service (QoS) guarantees, research in providing QoS support in multi-hop wireless networks has received much attention. In most cases, obtaining the information of end-to-end available bandwidth is the first step to provide QoS guarantees in multi-hop wireless network. This is due to that only with the information of available bandwidth we can do some QoS-aware operations, such as QoS routing, admission control and bandwidth reservation, and also we can sufficiently utilize the wireless resource. Furthermore, because of the strong dependence between available bandwidth and other QoS metrics, such as end-to-end delay, obtaining the information of available bandwidth also helps to provide systemic design for other QoS metrics. The purpose of this dissertation is to provide an accurate, systemic and scalable analysis approach for end-to-end available bandwidth estimation in multi-hop wireless networks.Every multi-hop wireless network is based on single-hop wireless links and multi-hop routes, which also appear in cooperative MIMO systems. We first improve the accuracy of available bandwidth estimation for single-hop wireless links, and then propose methodologies to estimate, model and predict the end-to-end available bandwidth for multi-hop routes. We will also quantitively analyze the bandwidth capacity of a typical cooperative MIMO system. In conclusion, the main contributions of this dissertation are as follows:We present a novel approach to improve the accuracy of the available bandwidth estimation in single-hop wireless link. The main innovation of this approach is that it differentiates the channel busy state caused by transmitting or receiving and that caused by sensing. This differentiation results in a more accurate estimation of the overlap probability of the idle times between two nodes and consequently a more accurate estimation of the available bandwidth between these nodes. Another challenge in available bandwidth estimation is to choose a proper estimation period. If the period is too short, the available bandwidth estimation samples will present high-frequency variations over the average value. On the other hand, if this period is too long, the estimated available bandwidth will not reflect the drastic change in the available bandwidth. To overcome this challenge, we present an approach based on a Kalman filter and a change detector. This approach achieves both good suppression of the high-frequency variations and fast tracking of the changes in the available bandwidth estimation.We propose and validate a methodology to analytically estimate the end-to-end available bandwidth of a given route in a multi-hop wireless network. To do this, we first formally present the calculation of the hidden node collision probability along a route. We then propose an optimization problem model, which takes into consideration the contending links'interference, hidden node collision and the possible channel bit rate difference between links in order to account for a realistic scenario. At last, we calculate the end-to-end available bandwidth via the optimization method. The proposed methodology is validated with extensive simulation experiments and shown to produce accurate results when applied in a wireless multi-hop network for both single-rate and multi-rate scenarios.We present a complete analytical model to provide throughput guarantees for applications in IEEE 802.11 based multi-hop wireless networks. The features of our model lie in that (i) it takes into account the realistic problems both from PHY layer and MAC layer in multi-hop wireless networks; (ii) it relates the network parameters to the QoS requirement of the flows; and (iii) it provides a what-if analysis, thus enables us to predict the performance of the network after a new flow with specific bandwidth requirement enters. Based on this model, iterative algorithms for end-to-end available bandwidth prediction are presented and validated. The proposed model is shown to be consistent with existing research efforts, yet it can give more precise quantitative analytical results.For the first time, we give a quantitive capacity analysis for cooperative MIMO technique. We divide the cooperative MIMO system into three subsystems that are amenable to well-known capacity theorems. After introducing the concept of"time efficiency"we integrate the composite system and derive the close form expression of its Shannon capacity. Based on this expression, we give some recommendations on when and how to use cooperative MIMO as well as on the optimal number of cooperative nodes. Based on the quantitive capacity analysis, we then derive an optimal resource allocation strategy and thus obtain the optimized end-to-end throughput of a multi-hop cooperative MIMO system. The strategy is obtained by modeling the problem to an optimization problem, which is solved by our mathematic derivation. The analysis gives the quantitive end-to-end capacity of the multi-hop cooperative MIMO system. It also shows that with an optimal resource allocation, the end-to-end throughput can be increased by 20% when compared with the equal resource allocation strategy.