| Summary: | Every year Telus invests tens of millions of dollars to increase the capacity of their
circuit-switched network in response to growth in demand. Growth has accelerated in
recent years due to rapid increases in data and Internet traffic, and the introduction of flatrate
long-distance services. This in turn has forced Telus to accelerate their investments
in the network to keep pace with demand. Circuit-switched technology is inefficient in
comparison to newer packet-switched technology, especially for data, which constitutes a
large and increasing portion of network traffic. In the near future Telus intends to begin
the move away from a circuit-switched to a packet-switched network. Investments in the
circuit-switched network will have very little value after the move to a packet-switched
network. Therefore Telus is looking for ways to reduce the investments needed in their
circuit-switched network.
One possibility for reducing investments is to use the existing network more efficiently
by improving the call routing. Every call has an origin switch and a destination switch.
Most calls are routed on a circuit linking the two switches directly. Call routing concerns
processing of calls where the direct circuit is either fully occupied or does not exist, in
which case calls are routed indirectly through one or more intermediate switches.
Presently Telus uses a method called Fixed-Hierarchical Routing. This is a fairly
inflexible scheme that routes overflow calls through one of only two of the 33 lower
mainland switches. The project team is investigating Sequential Alternate Routing which
would allow calls to be routed through any of the 33 switches. A key tool in this
investigation is a network simulator that was written for this project. The simulator will
be used to compare the performance of various new routing schemes, and the simulation
results will be used when recommending a scheme to Telus.
The simulator was subjected to a number of tests to verify that it functions as intended. A
key finding of those tests is that the simulator tends to underestimate the number of calls
blocked due to network congestion, but that this effect can be managed through the
judicious choice of the simulation run parameters. The tests also led to insights into the
choice of appropriate measurement criteria comparing the performance of routing
schemes.
|
|---|
