Adaptive Medium Access Control for Heterogeneous Wireless Sensor Networks

In this PhD thesis, we consider heterogeneous Wireless Sensor Networks (WSNs) in which several sensing devices with different characteristics coexist. In contrast to a homogeneous sensor network, in heterogeneous networks different sensors may sense different physical phenomena generating traffic that have different characteristics such as monitoring temperature, pressure, and humidity. Moreover, also information criticality can be heterogeneous. Deployment of nodes also introduces heterogeneity in the network. Depending on the specific application in fact, initial deployment can be random with the result that the distribution of nodes across the playground may be non-homogeneous. In addition to this, other factors such as node death because of energy resource exhaustion, mobility, or generic fault influence the heterogeneity of the distribution of nodes. All these characteristics can be considered as sources of heterogeneity of a WSN. Heterogeneity conditions may evolve during time and space, therefore, the design of a heterogeneous sensor networks requires adaptive mechanisms able to react to different characteristics, which is difficult to achieve. The goal of the thesis is to investigate the problems related to heterogeneity of sources in WSNs to design adaptive MAC methods that are able to take into account heterogeneity variations and are energy-efficient. We focus on two sources of heterogeneity. First, we study the problem of multiple traffic sources with different characteristics and constraints. Providing differentiated Quality of Service (QoS) such as low latency and high delivery ratio in large and multi-hops networks is a challenge due to limited energy resources of nodes. To solve this problem we propose an adaptive MAC protocol based on the asynchronous preamble sampling (PS) approach, a simple energy saving MAC technique, coupled with the idea of using a rendezvous time for data transmission. The proposed protocol (Low-Latency MAC, LA-MAC), is able to ensure efficient message forwarding throughout a multi-hop network thanks to the transmission of bursts. When messages need to be forwarded, each receiver behaves like coordinator to organize efficient transmission of contending senders. The innovation of the proposed protocol comes from combining enriched PS preambles to locally organize data transmission in a collision limited way. Extensive numerical simulations show that LA-MAC outperforms other state-of-theart protocols in terms of latency, delivery ratio, and energy consumption. The precise evaluation of the energy consumption in large wireless sensor networks that use preamble sampling MAC is difficult. In this thesis, we propose an analytic model for energy evaluation of PS that depends on the instantaneous traffic load of localized regions so that it is independent of the network traffic patterns that can also be heterogeneous. Second, we study dynamic WSNs with density of nodes that varies across space and time. Such networks are characterized by high variability in terms of node densities. Rapid density variation may affect the network state effecting the collision probability and energy consumption of devices. We address the case of dynamic wireless sensor networks in which i ii Abstract in French nodes and/or radio links may appear and disappear over time due to battery exhaustion, node mobility, or network management operations. With the work presented in this thesis we show that it is possible to provide QoS support in dynamic networks using an adaptive Density Aware MAC (DA-MAC) method. The proposed protocol offers a configurable channel sensing phase during which nodes request transmission opportunity in a way that avoids collisions. With DA-MAC, nodes periodically adapt their local protocol parameters to access the channel without collisions depending on local density state. The efficiency of the proposed protocol with respect to other state-of-the-art protocols is shown with extensive numerical simulations.