Abstract:
In Multi-radio multi-channel wireless mesh network architecture, a proper channel assignment
scheme must reduce interference to a minimum level while achieving maximum spectral
efficiency by means of channel reuse. Among other proposed schemes the Cluster-Based
Channel Assignment Scheme (CCAS) significantly improves the overall network performance.
CCAS logically partitions the network into non-overlapping clusters, while assigning a common
channel to all nodes within the cluster and using orthogonal channels for neighboring clusters to
prioritize the minimization of non-coordinated interference. Although CCAS significantly
reduces the complexity of minimizing non-coordinated interference, but a significant attention
has not been paid on channel assignment for inter-cluster links.
In our work we are enhancing this cluster based channel assignment by analyzing all aspects of
non-coordinated interference (i.e. FH, NH and IA) among inter-cluster gateway nodes, thus
maximizing the end-to-end throughput. Our analysis shows that for a given link, the interfering
FH and IA directional links are always reverse of each other. This leads us to an idea that
elimination of one type can automatically eliminate the other. In this way, while assigning
channel for a link, we have to focus only on two types of non-coordinated interfering links. We
have also computed the %age of interfering links for inter-cluster links and the possible
interfering area where the nodes reside creating that interfering links. We have then proposed an
inter-cluster channel assignment scheme on top of CCAS by utilizing 4 additional orthogonal
channels. The proposed scheme exploits the trade-off between numbers of inter-cluster links and
introduced non-coordinated interference. A sub-set of inter-cluster links is assigned orthogonal
channels while remaining inter-cluster links are not assigned any channel. The scheme results in
replacing certain inter-cluster links with the set of links have path length of up to 3. Simulation
results show an improvement of 40% for single flows on inter-cluster links. Results also show up
to 12% improvement in overall throughput for end-to-end flows.
