Optimising Relay Network Topology Using Generic Algorithm
The scheme of Relay wireless backpropagation presents a network optimisation model from centre to boundary. The center is connected by the DeNB base station via satellite transmission and serves as the connection and transmission of a series of RRN substations in its neighbourhood.
We implemented GA on the problem of RRN topology optimisation. This implementation was also optimised by multi-threads.
- Optimisation target: a sequence of sites allocated whether base station or substation, which is represented as a sequence of binary numbers.
- Value function: cost of constructing different types of stations and statellite communication expense.
- Generic Algorithmm: initialisation, network connection (greedy algorithm), value calculation, selection, crossover, mutation.
A generic framework synthesizes a group of classic methods:
- selection: Bolzmann selection and elite selection.
- crossover: uniform and/or crossover.
- mutation: mutation odds adjusted according to performance.