Symbiot: Congestion-driven Multi-resource Fairness

In this paper, we study the problem of multiresource fairness in multi-user sensor networks with heterogeneous and time-varying resources. Particularly we focus on data gathering applications run on Wireless Sensor Networks (WSNs) or Internet of Things (IoT) in which users require to run a serious of sensing operations with various resource requirements. We consider both the resource demands of sensing tasks, and data forwarding tasks needed to establish multi-hop relay communications. By exploiting graph theory, queueing theory and the notion of dominant resource shares, we develop Symbiot, a light-weight, distributed algorithm that ensures multiresource fairness between these users. With Symbiot, nodes can independently schedule its resources while maintaining networklevel resource fairness through observing traffic congestion levels. Large-scale simulations based Contiki OS and Cooja network emulator show the effectiveness of Symbiot in adaptively utilizing available resources and reducing average completion times.