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

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

8 The <strong>Design</strong> of Partially Survivable <strong>Network</strong>s 183<br />

The multi-period cellular design problem can be formally described as (M1)<br />

below, after defining parameters N, the set of cells; T , the number of time periods<br />

in the planning horizon; di j, the dem<strong>and</strong> from cell i in period j; ri j, the cost of<br />

connecting cell i to the root (MTSO) in period j; hi j, the cost of connecting cell<br />

i to the hub (backbone) in period j; si j, the cost of setting up a new link between<br />

cell i <strong>and</strong> the hub in period j; ti j, the cost of terminating an existing link between<br />

cell i <strong>and</strong> the hub in period j; k j, the capacity available at the hub in period j, <strong>and</strong><br />

variables xi j, which is 1 if cell i is assigned to the hub in period j <strong>and</strong>, 0 otherwise;<br />

ui j which is 1 if a new link is established between cell i <strong>and</strong> the hub in period j; <strong>and</strong><br />

vi j which is 1 if an existing link between cell i <strong>and</strong> the hub is terminated in period<br />

j. For notational convenience, xi0 is defined to be 0.<br />

min ∑ i∈N<br />

+ ∑ i∈N<br />

T<br />

∑ hi jxi j + ∑<br />

j=1 i∈N<br />

T<br />

∑ si jui j + ∑<br />

j=1 i∈N<br />

T<br />

∑ ri j(1 − xi j)<br />

j=1<br />

T<br />

∑ ti jvi j<br />

j=1<br />

s.t. ∑ di jxi j<br />

i∈N<br />

≤ k j j = 1,...,T (8.9)<br />

(M1) ui j − xi j + xi( j−1) ≥ 0 ∀i ∈ N, j = 1,...,T(8.10)<br />

vi j − x i( j−1) + xi j ≥ 0 ∀i ∈ N, j = 1,...,T(8.11)<br />

xi j<br />

∈ {0, 1} ∀i ∈ N, j = 1,...,T(8.12)<br />

ui j,vi j ≥ 0 ∀i ∈ N, j = 1,...,T(8.13)<br />

The objective is to minimize total cost, comprising the cost of connecting cells to the<br />

hub or the root in each period <strong>and</strong> the cost of establishing or terminating links from<br />

the cells to the hub in each period. Constraints 8.9 ensure that hub capacity is not<br />

exceeded in any period; constraints 8.10 <strong>and</strong> 8.11 ensure that the set-up costs <strong>and</strong><br />

termination costs, respectively are captured; constraints 8.12 enforce integrality of<br />

the assignment variables. Integrality of the set-up <strong>and</strong> termination forcing variables<br />

ui j <strong>and</strong> vi j need not be explicitly enforced — if the xi j are integral, these variables<br />

will be integral as well.<br />

Another formulation for this problem was introduced in Kubat <strong>and</strong> Smith [8].<br />

This formulation is amenable to column generation, <strong>and</strong> is presented as (M2) below.<br />

min ∑ fgρg<br />

g∈G<br />

s.t. ∑ aigρg = 1<br />

g∈G<br />

∀i ∈ N (8.14)<br />

(M2) ∑ b jgρg ≤ k j<br />

g∈G<br />

∀ j = 1,...,T (8.15)<br />

ρg ∈ {0, 1} ∀g ∈ G (8.16)

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

Saved successfully!

Ooh no, something went wrong!