Go to  Advanced Search

Dynamic subframe allocation in a reservation multiple access channel

Show full item record

Files in this item

Files Size Format Description   View
ubc_1993_fall_roorda_peter.pdf 4.665Mb Adobe Portable Document Format   View/Open
 
Title: Dynamic subframe allocation in a reservation multiple access channel
Author: Roorda, Peter D.
Degree Master of Applied Science - MASc
Program Electrical and Computer Engineering
Copyright Date: 1993
Abstract: Multiaccess protocols allow multiple geographically diverse users to communicate with a central station or each other over a single broadcast channel. Reservation protocols are a class of multiaccess protocols appropriate for satellite and radio networks. This thesis investigates dynamic time slot access control schemes for reservation multiaccess protocols similar to those investigated by Roberts [1] and others. Previous work has considered this model only for the case of a static frame subdivision, the majority of the work relying on approximate methods for analysis. In this thesis, schemes are considered where the subframe allocation is controlled on a frame by frame basis based on the current state of the finite user population. The controlled system is modelled as a Markov decision process with the subframe allocation as the decision variable. Optimality equations are provided and using infinite horizon dynamic programming techniques, the control policy that maximizes the average throughput on the channel based on complete state information is found. In addition to the optimal policy, we propose two heuristic dynamic allocation schemes that while being suboptimal, have implementation advantages over the optimal. Performance results indicate that the implementation of dynamic subframe allocation into the model significantly improves the performance of the network over previously considered fixed allocation schemes, both in terms of lowering the average message transmission delay and increasing the channel capacity. Proposed are methods of estimating the unknown state using a variation of Rivest's [2] pseudo-Bayesian scheme to allow approximation of the dynamic control schemes which depend on full state information. Simulation results indicate that the implementation of state estimation does not significantly erode the performance of the dynamic policies. Extending the model to incorporate multipacket messages is discussed and performance results indicate that in that case also, protocol performance may be improved using dynamic allocation.
URI: http://hdl.handle.net/2429/1435
Series/Report no. UBC Retrospective Theses Digitization Project [http://www.library.ubc.ca/archives/retro_theses/]

This item appears in the following Collection(s)

Show full item record

All items in cIRcle are protected by copyright, with all rights reserved.

UBC Library
1961 East Mall
Vancouver, B.C.
Canada V6T 1Z1
Tel: 604-822-6375
Fax: 604-822-3893