| dcterms.abstract | Recently, wireless local area networks (WLANs) have become an indispensable part of our daily life. To enhance the rate, range and reliability of WLANs, the IEEE 802.11n standard has introduced several new technologies, such as multiple input multiple output (MIMO), channel bonding and frame aggregation. However, the performance of WLANs is often unsatisfactory. In this dissertation, we study performance optimization in large-scale 802.11n WLANs, aiming at improving network throughput, reducing transmission failures, and enhancing the reliability and efficiency of link-layer multicast. In WLANs, clients need to associate with an access point (AP) to access the network. An AP and its associated clients operate on the same channel. The performance of WLANs can be adversely affected if too many clients associate with the same AP, or nearby APs operate on overlapping channels. The problem becomes more severe in 802.11n WLANs because of the channel bonding mechanism, which combines two adjacent $20$MHz channels together for data transmissions, and the presence of heterogeneous 802.11a/b/g/n clients. We first introduce mathematical models to estimate the client throughput in 802.11n WLANs. Based on these models, we propose AP association and channel assignment algorithms to maximize the network throughput. We further present low-complexity algorithms that minimize interference and contentions on high-rate clients in order to improve network performance. Another factor that affects performance in 802.11n WLANs is transmission failures, which are often caused by varying channel conditions. To ensure high reliability, failed frames are automatically retransmitted in WLANs. However, due to the temporal and spatial correlation of channel errors, retransmissions for a failed frame may also fail at a high probability. To address this issue, we design a cooperative retransmission protocol for 802.11n WLANs, where each node dynamically selects a neighbor that can overhear its transmission to help retransmitting. If the direct transmission fails, the selected neighbor retransmits the failed sub-frames of the aggregated frame. Transmission failures can also be caused by collisions, especially in WLANs that have heavy traffic load or a large number of clients. An AP can operate in the point coordination function (PCF) mode and poll its associated clients in turn, so as to achieve contention-free transmission. Nevertheless, the client polling in neighboring BSSs may still collide with each other due to the lack of coordination. Based on this observation, we study high-throughput collision-free client polling in large-scale WLANs that operate in the PCF mode. For many multimedia applications, e.g., video streaming, video conference, the sender needs to transmit every frame to multiple recipients, which could place tremendous traffic loads on WLANs. Link-layer multicast is a promising technology to greatly reduce this type of traffic loads thanks to the broadcast nature of the wireless medium. However, it is rarely used in practice due to the lack of reliability and efficiency. By taking advantage of smart antennas, we set up a system for multicast in 802.11n WLANs, with the objective to delivering multicast frames to all multicast clients reliably and efficiently. We have carried out extensive simulations and experiments, and the results show that the proposed schemes can significantly boost the network performance of large-scale 802.11n WLANs. | |
