CHALLENGES IN DIRECTIONAL ANTENNAS

Directional antennas have been proposed to improve the performance and capacity of Wireless MAC protocols for use in Mobile Ad hoc Networks (MANETs). Such antennas focus their beams in the direction of their receiver, allowing more of the signal power to be used in the direction of the transmission and simultaneously reducing the interference in other directions. However, several challenges and design issues arise that have no counterpart in omni-directional antennas – as a result, a large number of directional MAC protocols have been proposed. The thesis has three primary contributions organized into three parts: (1) It analyzes the directional MAC behavior in chain topologies, identifying some interesting interactions with upper layer protocols; (2) It identifies a head of line blocking problem in directional antennas and proposes new queuing policy to address it; and (3) It suggests a mechanism to passively discover directional neighbors and use that information to optimize multi-hop routes in directional antenna systems. In the first part, we study the interaction between a chain connection and an underlying directional MAC. The interaction of high level protocols with MAC in a MANET environment is often complex and unpredictable, resulting in unexpected behavior: such effects are well documented with Omni-directional protocols. Thus, we seek to understand whether similar interactions occur with directional antenna, iii especially when we consider the unique aspects of operation that directional antennas introduce. The problem characterization provides insight into how to design directional MACs to improve the performance of such connections. The second part identifies the inefficiency caused by FIFO Queuing mechanism while using directional antenna and proposes to use a different queuing policy which could take advantage of the channel utility factor provided by the underlying antenna system. Our results indicate that by using a greedy approach to schedule the packet which has the least wait time increases the overall throughput and end-to-end delay. The third part targets the problem of the inability of directional MAC protocols to discover directional neighbors (those neighbors reachable by a directional transmission but not by an omni-directional one). More specifically, directional antenna system use omni-directional mode to facilitate broadcasts and to find neighbors whose direction is unknown. The omni-directional mode is used by the routing layers to find the route. Even though the range of directional transmission is much larger than omni-directional, the routing layers are forced to route the packets through omnidirectional neighbors. The thesis proposes a new mechanism to update the routing layers with directional neighbors to find better routes. A directional DSR protocol is proposed and evaluated using such a mechanism. Preliminary results are encouraging, but comprehensive route maintenance and route error handling mechanisms needs to be added for the proposed protocol.