Gurashish Singh Brar

Computationally efficient scheduling with the physical interference model for throughput improvement in wireless mesh networks

Wireless mesh networks are expected to be widely used to provide Internet access in the near future. In order to fulfill the expectations, these networks should provide high throughput simultaneously to many users. Recent research has indicated that, due to its conservative CSMA/CA channel access scheme and RTS/CTS mechanism, 802.11 is not suitable to achieve this goal.In this paper, we investigate throughput improvements achievable by replacing CSMA/CA with an STDMA scheme where transmissions are scheduled according to the physical interference model. To this end, we present a computationally efficient heuristic for computing a feasible schedule under the physical interference model and we prove, under uniform random node distribution, an approximation factor for the length of this schedule relative to the shortest schedule possible with physical interference. This represents the first known polynomial-time algorithm for this problem with a proven approximation factor.We also evaluate the throughput and execution time of this algorithm on representative wireless mesh network scenarios through packet-level simulations. The results show that throughput with STDMA and physical-interference-based scheduling can be up to three times higher than 802.11 for the parameter values simulated. The results also show that our scheduling algorithm can schedule networks with 2000 nodes in about 2.5 minutes.

The SCREAM Approach for Efficient Distributed Scheduling with Physical Interference in Wireless Mesh Networks

It is known that CSMA/CA channel access schemes are not well suited to meet the high traffic demand of wireless mesh networks. One possible way to increase traffic carrying capacity is to use a spatial TDMA (STDMA) approach in conjunction with the physical interference model, which allows more aggressive scheduling than the protocol interference model on which CSMA/CA is based. While an efficient centralized solution for STDMA with physical interference has been recently proposed, no satisfactory distributed approaches have been introduced so far. In this paper, we first prove that no localized distributed algorithm can solve the problem of building a feasible schedule under the physical interference model. Motivated by this, we design a global primitive, called SCREAM, which is used to verify the feasibility of a schedule during an iterative distributed scheduling procedure. Based on this primitive, we present two distributed protocols for efficient, distributed scheduling under the physical interference model, and we prove an approximation bound for one of the protocols. We also present extensive packet-level simulation results, which show that our protocols achieve schedule lengths very close to those of the centralized algorithm and have running times that are practical for mesh networks.