|M.Sc Student||Taher Muhammad|
|Subject||MAC Protocol for Wireless Networks with Multipacket|
|Department||Department of Electrical Engineering||Supervisors||Professor Emeritus Adrian Segall|
|Dr. Gil Zussman|
|Full Thesis text - in Hebrew|
In multi-access wireless networks where a single channel is shared by a number of users, the Medium Access Control (MAC) is the scheduler. It schedules the transmissions of all users so that the common channel is efficiently utilized and the quality-of-service (QOS) requirement of each user is satisfied. The conventional assumption on such channels is that any concurrent transmission of two or more packets results in the destruction of all the transmitted information.
Contemporary networks challenge this assumption. Wireless networks today are able to successfully receive more than one packet when a number of packets is transmitted simultaneously. Therefore several improvements were added to the conventional assumption.
The first improvement is the capture effect and the second essential improvement added to the basic model is known as multiple-packet reception capability; according to this model, simultaneous transmitted packets can be successfully received by the receiver.
Until now, most of the studies focused on the impact of MPR on the performance of existing MAC protocols which were originally proposed for the conventional collision channel. The problem of designing protocols explicitly based on the general MPR channel model has rarely been touched. Nevertheless, fully utilizing the MPR capability is a nontrivial problem that calls for further study.
We introduce several approaches where the main idea is to split the users of the network into groups of “neighbors”. The users in each group are selected in a way that enables all uses in the same group to transmit simultaneously without destroying the transmissions of each other. Beside the grouping goal, the proposed protocols synchronize the transmission of the different groups.
The protocols utilize the capability of UWB networks to estimate the distance of the transmitter from the receiver, an estimate that is used as an input to the grouping algorithm. The grouping is done centrally in the base station.
We propose a protocol for saturated users in which the grouping stage is performed only once at network initialization. We also propose a protocol for non saturated users where the grouping is done in every frame. Both protocols are greedy in the way they build the groups.
In order to check our algorithm, we have developed a simulation program that allows us to compare our proposed protocols against the basic GDP. All our proposed protocols achieve improvement in the total performance of the network, improve fairness among the different users and prevent starvation.