아주대학교

초록/요약

In wireless networks with a large propagation delay, multiple access schemes should consider the effect of the propagation delay, as it is not negligible and affects network performance significantly. In terrestrial networks, a guard time length is very small owing to negligible propagation delay. On the other hand, in airborne networks and underwater networks, a guard time length is large to prevent collisions owing to large propagation delay. However, the guard time becomes large waste and degrades network performance such as link utilization, throughput, and packet delay. Therefore, reducing the guard time is very challenging in wireless networks with a large propagation delay. Before reducing the guard time, network topology should be considered because the performance of multiple access scheme is significantly affected by the network topology. In this dissertation, two types of networks are considered according to requirements derived from practical wireless networks with a large propagation delay. Firstly, a single-hop centralized network, in which all nodes are connected to one coordinator and transmit their packet to the coordinator through a randomly accessed channel, is considered. Secondly, a single-hop decentralized network, in which there is no central node and each node broadcasts a packet to all of the other nodes, is considered. For single-hop centralized networks, a random access scheme based on Slotted ALOHA (S-ALOHA) was studied. General S-ALOHA requires large guard time which causes lower normalized throughput. To reduce the large guard time, an existing scheme uses a time alignment mechanism based on the propagation delay. However, it assumed that propagation delay is known and did not consider acknowledgment. If the immediate acknowledgment is used, the effect of the propagation delay becomes twice. The efficient acknowledgment scheme is also required. Unlike the existing scheme, the proposed scheme does not require the assumption related to propagation delay. Instead, it has a framed structure starting with a beacon signal for supporting periodic propagation delay estimation. When all nodes randomly access a channel and transmit packet to coordinator, the guard time can be reduced based on the estimated propagation delay. Moreover, the beacon signal can also be used for acknowledgment. Numerical results from analytical model constructed show that the proposed scheme provides higher normalized throughput than the existing schemes. For single-hop decentralized networks, a distributed TDMA scheme was studied. In general, a large guard time is required to ensure broadcasting in TDMA networks with large propagation delay. This degrades link utilization. Thus, adaptive guard time control scheme and its MAC protocol are proposed to increase link utilization while ensuring broadcasting. For this, a time slot owner estimates propagation delay to its next time slot owner by exchanging timing information and adaptively adjusts its guard time based on that propagation delay. Moreover, the proposed schemes provide a distributed mechanism for propagation delay estimation and adaptive guard time control because there is no coordinator in decentralized networks. Numerical results and simulation results from analytical model and simulation model show that the proposed schemes provide higher link utilization than the existing scheme.