Decentralized Optimization for Multichannel Random Access

We consider schemes for decentralized cross-layer optimization of multichannel random access by exploiting local channel state and traffic information. In the network we are considering, users are not necessarily within the transmission ranges of all others; therefore, when a user is transmitting, it may only interfere with some users, which is different from most existing channel aware Aloha schemes. Besides, we also consider complicated traffic distribution, e.g. each user may choose to send packets to or receive packets from different users simultaneously. We develop decentralized optimization for multichannel random access (DOMRA). DOMRA consists of three steps: neighborhood information collection, transmission control of the MAC layer based on the instantaneous channel state information, and power allocation for each traffic flow on each subchannel. Simulation results demonstrated that DOMRA significantly outperforms existing channel aware Aloha schemes due to its exploitation of both multiuser diversity through cross-layer design and the inhomogeneous characteristics of traffic spatial distribution in the network. Besides, DOMRA performs closely to the globally optimum solution, which requires full network knowledge to be obtained. DOMRA can be applied to different types of wireless networks, such as wireless sensor networks and mobile ad hoc networks, to improve quality of service.