|
Student: Ngok Lam, Ph.D. student
Supervisor: Prof. Lorne G. Mason
Abstract: We study the capacity allocation problem in service overlay networks (SON)s with state-dependent connection routing based on revenue maximization. The dimensioning problem is formulated as two separate yet related optimization problems, namely the optimal routing problem and the optimal capacity allocation problem. The optimal routing problem is solved by employing an efficient event-dependent routing scheme that approximates the state dependent routing scheme. A call-admission-control (CAC) agent is included to allow reward sensitive connection GoS. This CAC mechanism provides an additional degree of control in the capacity allocation process and further improves the efficacy of the solution scheme. The optimal capacity allocation problem itself is solved by two different approaches; one employs the conventional gradient approach, the other uses the concept of average link shadow price that can be derived from information local to the links. The two approaches together with the CAC agent constitute four different capacity allocation schemes. They are tested and compared with one another. A key contribution of this paper is the extension of the shadow price method to solve a constrained capacity allocation problem efficiently. The second contribution is the illustration of the potential performance gains by allowing an additional degree of control in the optimization process.
[Full Description] [Paper (pdf format)]
|
Return to Service Overlay Networks Projects