| Concurrent Quality of Service Requirements:
A Comparative Analysis of Routing Path Selection Methods
Today’s modern network carries a wide range of traffic types (voice, video teleconference, messaging, and so on) which impose unique and concurrent Quality of Service (QoS) performance requirements. Traditionally, a single metric (such as latency or loss) is used as the criterion for picking the optimal routing path between the source and destination; this often results in other performance metrics being compromised. For example, the least end-to-end delay path may not be good from a packet loss perspective nor available bandwidth perspective.
A second method of selecting a routing path is using the minimum performance criteria. Under this method, a routing path is selected as long as concurrent multiple performance thresholds on all routing metrics are met regardless of how well they exceeded minimum requirements. For example, a routing path may be selected as long as the end-to-end delay is lower than 35 msec; the roundtrip packet loss is below .05% and jitter is less than 1 msec. No finer distinction is made with respect to any metrics on how superior their values are compared to the minimum requirements.
The third method is a variation of the second method where the selection criterion is picking the least cost path that meets the constraints of the other performance metrics. With this method, one may not end up with the absolute lowest cost path, but rather the lowest cost path that meets the other constraints. This can be implemented by finding the n lowest cost paths and then screening them for compliance with the other criteria.
The fourth routing method is built upon the Analytical Hierarchy Process (AHP). This routing method provides a compromise solution when there are competing performance requirements (loss, latency, throughput, security, and more) on routing paths of varying capabilities. The AHP is used to derive a single consolidated performance metric out of many performance metrics (latency, loss, delay jitter, bandwidth, security, and more) and using this consolidated metric’s value in selecting the optimal path. The essence of the AHP is to convert comparative evaluations of metrics on a priority scale to quantities using the eigen value and eigen vector concepts.
This paper will compare all the four methods of the route selection process with several examples and make a recommendation. |
|
Dr. Subramaniam Kandaswamy is currently a senior research staff at the Johns Hopkins University, Applied Physics Laboratory. He has over 18 years of experience in: Performance Analysis, Statistics, Simulation & Modeling, Network Management, and Systems Engineering. He has hands-on expertise in data (IP, ATM, and Frame) and voice networks. He is currently working with a team supporting the OASD-NII in developing NCID Network and Mission Performance Modeling. He has presented papers at several national conferences and symposiums.
|