Open Collections

A study of bandwidth allocation mechanisms on sonet-ring-based transport networks

University of British Columbia

A new SONET-based network simulator was designed and implemented to study the performance of different routing mechanisms on transport networks based on the SONET ring architecture. The simulator incorporates some unique features of the SONET technology. Three static routing schemes were evaluated on the simulator at different levels of traffic load. The hop-by-hop routing (HHR) scheme uses the physical network as a fat bit pipe to transport the data traffic from one node to its neighboring node. Utilizing the Digital Cross-connect System in the SONET technology, virtual topology routing (VTR) schemes establish the multiple-hop point-to-point links across a SONET ring in order to reduce overall nodal processing time. The simulation results show that virtual topology routing schemes have advantage of low network latency over the hop-by-hop routing scheme only when the bursty traffic load is well below the network capacity. In medium to high traffic load conditions, increased queuing delay counteracts the reduced overall nodal processing time in the VTR schemes, resulting in high network latency, while the HHR scheme have the advantage of high bandwidth utilization and low packet-drop rates. To overcome the drawbacks of the static routing schemes, dynamic bandwidth-allocation mechanisms combining VTR and HHR principles were suggested. The mechanisms allocate bandwidth dynamically between VTR and HHR regions in response to the changes in traffic patterns on a SONET ring. Both centralized model and distributed model were proposed. For the centralized model, the problem of optimal effective bandwidth allocation on the Unidirectional Path Switched Ring (UPSR) architecture was formulated and a solution based on a Greedy algorithm with cost of 0(n²m + mlogm) was provided for UPSR architecture with n nodes and m flows. A heuristic bandwidth optimization algorithm based on the solution for UPSR was developed and evaluated numerically for Bidirectional Path Switched Ring (BPSR) architecture. The possibilities of extending the heuristic algorithms to other SONET-ring-based architectures were also explored. For the distributed model, a mechanism was proposed to maintain the tension between bandwidth utilization and network latency in local scope rather than in global scope. By resolving problem early and locally, intensive computation of global bandwidth optimization and synchronization of global reconfiguration may be avoided.

Thesis/Dissertation

application/pdf

2009-06-15

For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.

http://hdl.handle.net/2429/9172

Computer Science

University of British Columbia

UBCV

DSpace