|
Researchers: Ngok Lam, Ph.D. Student & Prof. Lorne Mason
Description:
The key components of Service Overlay Networks (SON) are the SON gateways and the interconnecting logical links that lie on the top of one or more physical links. SON gateways can be treated as routers that relay service specific data and perform control functions. SON logical links provide connectivity to the SON network through
existing physical links. The SON gateways are connected to adjacent gateways by the logical links. To provide SON service, the SON provider has to purchase bandwidth and QoS guarantees from the corresponding network infrastructure owners via Service Level Agreements (SLA). It can be easily noticed that the optimum amount of capacity to be purchased from the infrastructure owners so as to maximize the net revenue is a big challenge faced by the SON providers.
In this study, we dimension the SON network based on revenue maximization, in
this aspect we are not only considering the net income in our objective function, but
moreover we are also incorporating the idea of average network shadow price in the
dimensioning process to reflect the sensitivity of net revenue to the dimensions of the
links. We consider the SON network as a generic network and provide a framework
for dimensioning based on the traffic rewards. The dimensioning problem was
formulated as a constrained optimization problem with respect to two routing models.
From the KKT conditions of the optimization formulation, we devise an iterative
method that leads to the near-optimal solutions. Compared with the previous studies
in the literature, our models allow more flexible routing formulation such that each
path can compose of arbitrary number of links. We also incorporate two sophisticated
routing schemes to better approximate the state dependent routing scheme assumed in
the SON environment. In addition, we are not only presenting the optimization
models theoretically, we also have detailed discussions on the implementation issues
of the models and performed numerical studies that verify the models. One major
novelty of our study is that we provide an economic integration of the control layer
and the dimensioning layer of the SON network through the use of average shadow
price concept.
Publications:
N. Lam, Z. Dziong and L.G. Mason, Network capacity allocation in service overlay networks, in Proc. International Teletraffic Congress (ITC), Ottawa, ON, Canada, June 2007. [Paper (pdf format)]
|