|M.Sc Student||Tcharny Galina|
|Subject||Fuzzy Group Communication in Mobile Ad Hoc Networks|
|Department||Department of Computer Science||Supervisor||Professor Roy Friedman|
Mobile Ad-Hoc Networks (MANET) are networks of mobile devices that are formed in an ad-hoc manner. That is, the devices that participate in such networks have wireless communication capabilities with a limited range transmitters, and therefore can directly communicate with other devices in their range. Some of the devices occasionally volunteer to forward some of the messages they receive, or in other words, act as routers, thereby forming a network. In this thesis we investigate the performance of four important issues in distributed computing - failure detection, stability detection, flow control and buffer management in ad- hoc networks. In ad-hoc networks, short lived disconnections and network partitions are common due to their mobility. Nevertheless, with existing approaches for reliable multicast and group communication, even a single disconnected device may slow down all other devices. For example, in group communication, in order to deliver messages reliably, each member usually maintains copies of all messages it sends and receives for potential retransmission. A message that has been received by every member in the group can be safely discarded. In ad-hoc networks, the latency of collecting information from all members may become very high due to possible disconnections.
Hence, a single disconnected node may cause message buffer overflow at all other group members.
We propose an infrastructure that allows one to detect such problematic devices, and enables the others to make progress despite of short lived disconnections. More specifically, we propose the notions of fuzzy membership and fuzziness level of a group member, and utilize them in order to improve the performance of group communication and reliable multicast despite of short lived disconnections.