Channel Allocation in
Cellular Networks
Heuristic Methods
Jie Chen David Seah Wen Xu
Overview
● The Channel Allocation Problem
● Three Heuristic Algorithms
● Simulated Annealing
● Genetic Algorithm
● Tabu Search
● Comparison and Observations
● Conclusion
The Problem
● Channel Allocation Problem (CAP)
● Traffic Vector
● T = [32, 26, 14, 32, 18, 20, 14]
● Decision Variable
The Problem
● Interference Matrix
● Co-channel constraint
● Co-cell constraint
2 1 0 0 0 0 0
1 2 1 1 1 0 0
0 1 2 0 1 0 0
I= 0 1 0 2 1 1 0
0 1 1 1 2 1 1
0 0 0 1 1 2 1
0 0 0 0 1 1 2
● Cost function
The Problem
● Neighborhood definitions
● Random Selection
● Cycling
● Probability-based
● GSAT-like
Simulated Annealing
● Neighborhood Definition
● Random Selection
● Initial Temperature
● Pinitial = exp(-Δ cost / T0 )
● T0=k σ, where k= - 3/(ln Pinitial)
● M, b, a
Genetic Algorithm
● Fitness Mapping
● Selection
● Crossover
● Mutation
● Standard
● Greedy
● Population Size
● Elitism
Tabu Search
● Neighborhood definition
● GSAT-like
● Tabu Restriction
● A swap pair changes from (0,1) to (1,0)
● A swap pair changes from (1,0) to (0,1)
● At least one of the above occurs
● Tabu Tenure & Candidate List Size
Tabu Search
● Memory usage
● Short term
● Aspiration Criterion
● Global Aspiration by Objective
● Tabu list data structure
● Three-dimensional 7*50*50 array
Comparisons / Observations
● Results
Cost Eval. SA GA TS GS
Avg. 10,000 81.3 114.13 81.6 90.5
Best
50,000 77.5 87.38 77 89
Cost
100,000 76.8 84.25 76.6 89
Std 10,000 1.49 3.44 1.27 4.68
50,000 0.85 2.22 0.95 3.29
100,000 1.32 2.15 0.84 3.29
Best Cost Found 75 78 75 85
Comparisons / Observations
● Results
Comparisons / Observations
● Result Summary
● SA and TS performed comparably
and better than GA
● GA performed better than Greedy Search
(GS)
Comparisons / Observations
● How good are the results acquired?
● Compared to GS and RGS
● What is the global optimum?
● Lower bound is 63
● How many users can actually be
supported?
● Upper bound is 133
● Lower bound is 130 (priority-based optimization
method)
Conclusion
● Three major heuristic methods (SA, GA,
TS) have been applied to CAP for
cellular networks.
● TS and SA can achieve good
performance, but GA performs worse.
● A simple priority-based optimization
method is presented to find the optimal
channel assignment.