Scribd
Upload
141 views

HW 1 Soln

This document contains solutions to homework problems on network design. It addresses problems related to formulating link-demand-path models with different constraints, optimizing average network delay, determining feasible flows under ECMP routing rules, and expressing the shortest path problem as a linear program. The solutions demonstrate how to rewrite formulations by removing demands, define new objective functions, analyze feasibility conditions, and model routing problems mathematically.

Uploaded by

gooseneck06
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
141 views

HW 1 Soln

This document contains solutions to homework problems on network design. It addresses problems related to formulating link-demand-path models with different constraints, optimizing average network delay, determining feasible flows under ECMP routing rules, and expressing the shortest path problem as a linear program. The solutions demonstrate how to rewrite formulations by removing demands, define new objective functions, analyze feasibility conditions, and model routing problems mathematically.

Uploaded by

gooseneck06
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Network Design Homework 1 Solution

CSCE 5530
Department of Computer Science
University of North Texas
September 29, 2009

1 Exercises
1.1 Ex. 2.1 [2]
Consider the formulation (2.1.2). Now suppose there is no demand between
node pair 1 and 3, i.e., ĥ13 = 0. Rewrite the complete formulation for this
modified problem in link-demand-path-identifier-based notation.
Solution Link path formulation in link-demand-path identifier based no-
tation
When ĥ13 = 0 in (2.1.1), then the paths for this demand do not need to
be considered. Thus, the formu- lation is:
minimize
F = x1 1 + 2x1 2 + x2 1 + 2x2 2
subject to (constraints)
x1 1 + x1 2 = 5
x2 1 + x2 2 = 8
x1 1 + x2 2 ≤ 10
x1 2 + x2 2 ≤ 10
x1 2 + x2 1 ≤ 15
x1 1, x1 2, x2 1, x2 2 ≥ 0.

1
1.2 Ex. 2.3 [2]
Identify another goal that can be of interest in a network and formulate the
corresponding objective function for the three-node example given by con-
straints (2.1.2). Determine the optimal solution.
Solution
Another objective for (2.1.2)
An important objective for a network is average network delay (see Model
(7.1.7) in Chapter-7 for the general formulation and derivation). Specifically,
here our objective will be:
x̂12 +x̂123 +x̂213 x̂132 +x̂13 +x̂213 x̂132 +x̂123 +x̂23
minimizeF = 10−x̂ 12 +x̂123 +x̂213
+ 10−x̂ 132 +x̂13 +x̂213
+ 10−x̂ 132 +x̂123 +x̂23

1.3 Ex. 2.6 [2]


Consider the example depicted in the figure 2.6 with link capacities altered
as follows: all link capacities are equal to 1 except for link 3-5 which has
capacity 1 12 and links 3-4 and 4-5 which have the same capacity 12 . Does a
weight system exist that induces a feasible solution under the ECMP rule.
Solution
For the considered set of link capacities and demand volumes, there is no
weight system that induces feasible ECMP flow in this network. On the
other hand, feasible flow allocation to paths is possible (but without any
weight system). Such a feasible flow is for example obtained by assigning
flow x11 = 1 to path 1 → 3 → 5 → 7, flow x21 = 12 to path 2 → 3 → 5 → 6,
and flow x22 = 12 to path 2 → 3 → 4 → 5 → 6.

1.4 Shortest path problem


Write the shortest path problem in linear programming style. I.e., you want
to find the minimum length path from s to t amongst all the candidate paths
from s to t, etc.
Solution [1]
Given a graph G = (V, E), with weight function w : E → R mapping edges
to real-valued weights, a source vertex s, and a destination vertex t. The
Bellman-Ford algorithm computes the value d[t], which is the weight of a
shortest path from s to t. This problem is expressed as a linear program
determining a set of variables and constraints that define the shortest path

2
from s to t. At Bellman-Ford algorithm termination, algorithm has computed
for each vertex v, a value d[v], such that for each edge (u, v) ∈ E, we have
d[v]≤ d[u]+w(u, v). The source vertex receives a value d[s]= 0, which is
never changed. Thus we obtain the following linear program to compute the
shortest-path weight from s to t.
maximize d[t]
subject to
d[v]≤ d[u]+w(u, v) for each edge (u, v) ∈ E
d[s]= 0
In this linear program, there are kV k variables d[v], one for each vertex
v ∈ E. There are kEk + 1 constraints, one for each edge plus the additional
constraint that the source vertex always has the value 0.

References
[1] T. H. Cormen, C. E. Leiserson, R. L. Rivest, and C. Stein. Introduction
to Algorithms, Second Edition. McGraw-Hill Science/Engineering/Math,
July 2001.

[2] M. Pióro and D. Medhi. Routing, Flow, and Capacity Design in Com-
munication and Computer Networks. Morgan Kaufmann Publishers Inc.,
San Francisco, CA, USA, 2004.

You might also like