16.11.2012 Views

Wireless Network Design: Optimization Models and Solution ...

Wireless Network Design: Optimization Models and Solution ...

Wireless Network Design: Optimization Models and Solution ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

178 Syam Menon<br />

with the MTSO. Every cell has to be connected to the MTSO, either directly or<br />

through hubs that aggregate traffic from multiple cells <strong>and</strong> connect to the MTSO<br />

via higher capacity links. With increasing size, the latter approach becomes more<br />

attractive as it enables the owner of the network to reduce costs by taking advantage<br />

of economies of scale.<br />

The increased efficiency comes with reliability concerns however, as the failure<br />

of even one high capacity line could have a significant impact on the entire network.<br />

Common approaches to tackling this issue include the use of multiple paths (Cardwell<br />

et. al. [3]) <strong>and</strong> self-healing SONET rings (e.g., Sosonsky <strong>and</strong> Wu [17]). SONET<br />

rings are particularly convenient as they provide high capacity at a reasonable cost.<br />

In addition, if the cable is severed at some point in the ring, data can re-route itself<br />

with little to no loss in information. The increased flexibility <strong>and</strong> b<strong>and</strong>width availability<br />

of SONET provides significant advantages over older systems. These include<br />

reduced equipment requirements, increased network reliability, <strong>and</strong> a synchronous<br />

structure which greatly simplifies the interface to digital switches <strong>and</strong> digital adddrop<br />

multiplexers.<br />

Most large companies lease transmission capacity on their SONET rings to other<br />

organizations. Dutta <strong>and</strong> Kubat [4] introduced the problem of designing partially<br />

survivable interconnect networks for a cellular system that has access to this leased<br />

capacity. They represented the problem as an integer program that minimizes the<br />

cost of transmission subject to reliability requirements <strong>and</strong> assumed a self-healing<br />

ring topology for the backbone network. They showed that the problem was NPhard<br />

<strong>and</strong> developed an efficient heuristic based on Lagrangian relaxation. This was<br />

further addressed in Menon <strong>and</strong> Amiri [13], who showed that the problem could be<br />

solved efficiently via dynamic programming. This single-period problem is the first<br />

one addressed in this chapter.<br />

Cellular telecommunications systems tend to be more flexible than traditional<br />

ones. As a result, traditional approaches to telecommunications network design are<br />

often inappropriate for the design of cellular networks, <strong>and</strong> approaches which explicitly<br />

incorporate the increased flexibility into the design process are necessary.<br />

The second problem we consider in this chapter is a multi-period model that was<br />

addressed in Kubat <strong>and</strong> Smith [8] <strong>and</strong> Menon <strong>and</strong> Amiri [14]. It exploits the flexibility<br />

of cellular systems by designing a multi-period cell-MTSO interconnection<br />

network. The objective is to minimize the total interconnection costs over the planning<br />

horizon, while determining homing decisions for the cells in each planning<br />

period.<br />

The third problem considered in this chapter was introduced by Kubat et al. [9].<br />

In this problem, reliability requirements are incorporated in the face of capacity constraints.<br />

They represent the problem as an integer program that minimizes the cost<br />

of transmission subject to reliability requirements while assuming a tree topology<br />

for the underlying network. We introduce a new formulation for this problem <strong>and</strong><br />

present a procedure based on column generation to solve it.<br />

Each problem is discussed in a separate section of this chapter. The single period<br />

problem is addressed in Section 8.2, while the multi-period <strong>and</strong> capacity constrained

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!