Cross Layer Protocols and Architecture for Wireless Sensor Networks

Abstract

In the recent few years, a new class of networks called Wireless Sensor Networks (WSN) that expand human ability to monitor and interact remotely with the physical world has appeared. These networks consist of smart sensor nodes that collect huge amount of hitherto unknown data from the environmental, process it locally, and send the information to central Master Station (MS). On receiving the data, MS analyzes the information therein and initiates suitable response, if required. Typical applications of WSN include, but are not limited to environmental mon- itoring for precision agriculture; animal habitat monitoring; climate monitoring; in- dustrial machine control and monitoring; Fire tracking and assistance to Fire Fighters; tracking movement of a source (human, animal, object); civil infrastructure monitor- ing etc. Nevertheless, to exploit the full potential of sensor networks, it is First required to address the peculiar limitations of these special networks and the resulting techni- cal issues. For example, in WSN, nodes work under severe resource constraints like limited computational, storage, energy, and communication bandwidth capabilities. The primary goal of protocols designed for sensor networks has been energy efficiency so as to prolong the network lifetime and make WSN applications econom- ically feasible. With the advent of new sophisticated applications of sensor networks, WSN users need protocols that are self organized; self healing; fault tolerant; latency efficient; have predictable performance; can adapt to topology changes and varying traffic requirements. The protocols should be very simple to implement on memory and computationally constrained nodes. To cope up with the complexities introduced by computational; storage; power; and communication bandwidth capabilities of the sensor nodes at one end and the design needs of versatile applications at the other, cross layer approach is used in the protocols proposed in the thesis. The First protocol proposed is Self Organized, Flexible, Latency, and Energy Efficient (SOFLEE), a Medium Access Control (MAC) and routing based cross layer protocol for WSN. SOFLEE uses Time Division Multiple Access (TDMA) based MAC that combines routing information during time slot allocation for uninterrupted data forwarding. TDMA guarantees fairness and avoids major energy waste arising from idle listening, collision, and overhearing. Time slot allocation is done centrally by MS to provide a collision free and fair media access. MS allocates same transmission slot to nodes that are two hops apart to increase channel spatial reuse and decrease data latency. Unlike a conventional TDMA based protocol, SOFLEE provides exibility to transfer data slots among nodes and priority based slot scheduling to adapt to dynamic traffic patterns of applications. For data gathering at MS, SOFLEE uses parenthood willingness to forward data to MS through unidirectional tree rooted at MS. Parenthood willingness of a node is decided using: (i) its location with respect to MS, to forward data in correct direction; (ii) its number of children, to prevent local congestion; (iii) its residual energy, to uniformly distribute energy load of being a parent node and (iv) its parent child communication link reliability to guarantee consistent data delivery. Finally, simple, memory, and energy efficient techniques for: (i) hop by hop congestion control; (ii) catering to orphan nodes, link breakdowns, and node deaths increases robustness of SOFLEE. The second protocol proposed is Low Energy Fuzzy based Unequal Clustering Multihop Architecture (LEFUCMA). It is a cross layer hierarchical clustering protocol that encompasses neighbour finding, cluster head selection, clustering, and routing protocols. The neighbour Finding protocol uses localized information to organize the network into a sectored layers structure. Most real world applications based WSN are deployed on a large scale and inevitably posses uncertainties like unstable channel conditions and varying network topology. Under these conditions use of control systems like fuzzy logic which can tolerate the imprecise and noisy inputs for protocol development can give best results. Hence, LEFUCMA uses fuzzy logic with residual energy, number of neighbouring nodes (to ensure minimum energy for intra cluster communication), packet reception rate (to ensure reliability) and distance of node from MS (to ensure minimum energy for inter cluster communication) as fuzzy descriptors for cluster head selection. For a multihop clustered network the cluster traffic is more in cluster heads in the high node density area and relay traffic is more in the cluster heads near the MS (as they are required to relay packets of all distant cluster heads). To evenly distribute cluster and relay traffic, LEFUCMA again uses fuzzy logic with node density and distance of area from MS as fuzzy descriptors to decide the number of cluster heads in a given area. To further limit the relay traffic of cluster heads near the MS, LEFUCMA uses an unequal clustering mechanism which partitions the nodes into clusters of unequal size, with clusters closer to the MS having smaller sizes than those farther away from the MS. Unequal clustering balances energy consumption of cluster heads and also solves the hot spots problem that occur in multihop clustered network. Finally, a simple and energy aware multihop routing protocol for inter cluster communication and final data delivery to the MS is proposed. The inter cluster routing protocol decides the next hop cluster head considering its: (i) residual energy; (ii) distance from MS and from current cluster head (represents energy required for communication); (iii) number of cluster members (represents intra cluster traffic); (iv) number of descendant cluster head nodes for which it is already a relay node (represents inter cluster traffic). The third protocol proposed is Fuzzy and Ant Colony Optimization based com- bined MAC/Routing cross layer protocol for WSN (FAMACROW) that incorpo- rates cluster head selection, unequal clustering, and inter cluster routing protocols. FAMACROW uses fuzzy logic with residual energy, number of neighbouring nodes, and quality of communication link as input variables for cluster head selection. To avoid hot spots problem, FAMACROW uses an unequal clustering mechanism with clusters closer to MS having smaller sizes than those far from it. Finally, Ant Colony Optimization technique is used for reliable and energy efficient inter cluster multihop routing from cluster heads to MS. The inter cluster routing protocol decides relay node considering its: (i) distance from current cluster head and that from MS (which represents energy required for communication) (ii) residual energy (for energy dis- tribution across the network) (iii) queue length (for congestion control) (iv) packet reception rate (which represents the reliability of communication link). Finally, the research proposes Fuzzy Cross (FUCR), a novel cross layer decision making and information sharing architecture that enables protocols to achieve energy efficiency, reliability, and low data latency. For this, it does not change the underly- ing/existing protocols at different layers. Instead, it provides an administrative plane that collects: (i) current state information of the node like: its residual energy; sensi- tivity of sensing unit and (ii) local view of the network like: successful data delivery; channel assessment and packet dispatch rate. A fuzzy agent uses this information as input descriptors for running fuzzy logic. Its output is used by physical layer protocol to decide transmit power of node; by data link layer protocol to decide retransmission time out, back off time and duty cycle; network layer to determine prospect of the node to become a relay node and application layer to determine prospect of node to become a reporting node. To summarize, this thesis describes the design, implementation, and evalua- tion of cross layer protocols and architecture for WSN that focus on achieving energy efficiency, provisioning predictable performance, and supporting reactivity and adapt- ability, with constantly changing environment. The main contributions of this thesis are: A Self Organized, Flexible, Latency, and Energy Efficient Protocol that en- sures: fairness; energy efficiency; adaptability to traffic pattern; higher channel utilization; throughput; and lower latency, while respecting the computational, memory, and power constraints of the nodes. Low Energy Fuzzy based Unequal Clustering Multihop Architecture that en- sures: energy efficient cluster head selection; avoiding hot spots problem; provid- ing simple, energy efficient, and scalable inter cluster routing for data delivery to MS. Fuzzy and Ant Colony Optimization based combined Medium Access Con- trol/Routing cross layer protocol that ensures energy efficient cluster head se- lection; avoids hot spots problem and provides simple, reliable, energy e cient, and scalable routing protocol for inter cluster communication. Fuzzy cross, a simple, generic, extensible, exible, portable, and stable decision making and information sharing architecture for WSN that enables the protocols to achieve energy efficiency, reliability, and low data latency.