Fair Buffer Allocation Scheme for Integrated Wireless Sensor and Vehicular Networks using Markov Decision Processes

In sparsely deployed Wireless Sensor Networks (WSNs) the end to end connectivity between sensor nodes and the sink is not always available. To allow communication between sensor nodes and the sink, mobile vehicular nodes often come into play to collect data from isolated nodes, temporarily store it in their buffer and ultimately deliver to the sink. In these store and forward networks, buffers at the roadside relay nodes become critical resources for overall network performance and need to be allocated fairly among sensor nodes. In this paper we have proposed a fair buffer allocation policy for roadside relay nodes in an integrated wireless sensor and vehicular network scenario. We have considered a network with sparsely deployed sensors for gathering desired information from surroundings which is ultimately delivered to the sink via vehicular nodes. In particular, we first present the model of the buffer at the road side relay node and then formulate constrained semi-Markov decision process (SMDP) for fair buffer management. The SMDP provides an optimal decision policy to achieve fair buffer allocation at the roadside relay nodes. The model is flexible and can be realized using look up table approach. Performance evaluation results reveal the significance of the proposed policy in achieving the fair buffer allocation for different number of buffer states as well as data arrival rates.