Predictable, Scalable QoS Routing for Ad Hoc Wireless Networks based on Heavy-Tailed Statistics
Agency / Branch:
DOD / ARMY
In this proposal, we identify QoS metrics based on our insights on heavy-tailed phenomena in ad hoc wireless networks, determine the potential performance impact of heavy-tailedness on ad hoc routing, define routing policies based on such insights and metrics, and propose a QoS Routing protocol (HAQR) that captures the heavy-tailed nature and provides quality of service for multi-hop wireless ad hoc networks. Performance metrics are identified as the 1st and 2nd order statistics of throughput, delay and packet loss. With respect to heavy-tailed traffic, L1 norm based estimation methods are exploited for the metrics estimation. The unique predictability nascent in heavy-tailed statistics is utilized to classify short- and long-lived flows and predict into the future for routing optimality and stability. In addition, time scales need to be carefully selected to ensure routing stability. HAQR possesses the advantages of both proactive and reactive routing paradigms and it has the following desirable properties: it (1) captures the heavy-tailed based metrics for routing decision, (2) adapts to the dynamic environment and accommodates the imprecise topology information, (3) reserves and releases bandwidth efficiently, and (4) reduces initial latency and routing overhead. The proposed algorithms for heavy-tailed based QoS routing are predictable, efficient and scalable.
Small Business Information at Submission:
Contracts and Proposals Manager
Research Institution Information:
Intelligent Automation, Inc.
15400 Calhoun Drive, Suite 400 Rockville, MD 20855
Number of Employees:
Hovde Hall of Administration, 610 Purdue Mall
West Lafayette, IN 47907
Nonprofit college or university