Maximizing throughput while maintaining fairness and priority in wireless ad-hoc networks

Abstract

Ad-Hoc wireless network, not a new field of research, has been gaining a lot of popularity in recent years because of its applications in emergency rescue, surveying, military, sensor networking, entertainment and community wireless networking etc. A lot of research has been done on improving the performance and capacity of ad-hoc networks in that period of time. However, not much has been done in the analytical modeling of inter nodal interference and its effect on transmission patterns of wireless ad-hoc network. In our work we investigated and developed a contention based ad-hoc network to study the interference impact on network performance. In our model we represented these inter nodal conflicts using 'Conflict Graph or Conflict Matrix' and presented scheduling schemes to address issues like network throughput, fairness and QoS (priority) in a network. We particularly studied the transmission patterns in a network given a specific conflict graph which in turn depends on factors like number of nodes and their placement, area of coverage, type of antenna (unidirectional or omnidirectional) etc. We designed five scheduling schemes in order to provide solutions to issues like maximizing network throughput, fairness and priority using the concept of 'Transmit Groups', which are groups of nodes that are capable of transmitting together without any interference. For simulations, we have assumed a central entity which controls and schedules packet transmission from nodes. Using MATLAB simulations, we analyzed the throughput in hundreds of sample networks which differ based on number of nodes, number of conflicts and number of high priority nodes in the network. With the help of these results we not only proved that interference is an important factor in performance of wireless ad-hoc networks , but also its impact on different aspects of network depends on the number of nodes and number of conflicts. Based on the length of longest transmit group in a network and its variation with number of nodes, we calculated the maximum throughput possible. In our QoS schemes, we compared the success rate of achieving a pre assumed relative throughput by S-D pairs.