Dissertation Proposal

Fei (Sophie) Che

Fourth Year Ph.D. Candidate @ UDel CIS

We propose to study two distinct aspects in network optimization for specific requirements of different types of networks:

In Part I, we consider optimization related to
connectivity in wireless ad hoc networks. Wireless Ad hoc Networks are networks where wireless nodes cooperate to dynamically form a network without necessarily using any pre-existing infrastructure. A key goal in managing such a network is to prolong the network lifetime while preserving the network connectivity. We propose to work on two approaches to connectivity problems. In the first approach, we will consider topology control problems which utilize node power assignments so as to maintain a connected network. We will focus on the Mobile Topology Control problem
in
Mobile wireless Ad hoc Networks (MANETs) where nodes are mobile. Such problems have not yet been studied well from a theoretical viewpoint and the major challenge comes from accommodating the node movement. The second approach that we will consider is to introduce a small number of more costly but more powerful relay nodes so as to communicate with network nodes and with other relays. The Relay Node Placement problem we propose to study is to give a placement of relays aiming at some particular objective, such as minimizing the number of relays, maximizing the number of connected sensors, etc.
The
replacement of relay nodes will also be studied.

In Part II we consider optimization for
multicast channelization in large scale data dissemination networks. Such networks consist of a large number of information flows and information users, where individual users are interested in {\it specific (not all!)} flows. Instead of the costly broadcasting, multicasting is usually used in such networks. The  channelization problem is to construct multicast groups in a way that the network cost is substantially less in comparison with using a single multicast group. In our proposed research, we first consider Incremental Channelization where the channelization instance can change over time. The goal is to compute a good solution based on existing results
in significantly less time than to compute ``from scratch''. We then consider
Rendezvous Point Selection where multicast groups have been established based on user preferences for a set of available flows. This problem aims at minimizing the total network traffic without overloading any single host node.

Network Optimization

Publications

¡¤ F. Che and E. L. Lloyd, "Incremental Channelization", MilCom¡¯09 [pdf]

¡¤ F. Che and E. L. Lloyd, "The complexity of RP selection in Multicast Channelization", MilCom¡¯09 [pdf]

¡¤ F. Che, E. L. Lloyd and L. Zhao, "Improved topology control algorithms for Simple Mobile Networks", AdHocNets¡¯09 [pdf]

¡¤ F.Che, L.Zhao and E.L. Lloyd, "Optimal Relay Node Fault Recovery", IWASN¡¯07 [pdf]

¡¤ F. Che, ¡°Integrated Positioning Algorithm Using Multiple Satellite Navigation and Positioning Systems¡±, Radio Engineering of China (Journal), Vol. 37, No. 3, 2007, Pages 34-35.