This section is devoted to discuss the main approaches used to design the MAC and routing and their cross layer protocols in order to point out their advantages and disadvantages in the domain of WSNs and to provide a motivation for this thesis. In terms of the MAC protocols, it can be seen that CSMA-CA as a distributed mechanism with low overhead is appropriate for most of the WSNs applications, i.e., CSMA-CA does not require a node to be equipped with a special transceiver (as the case in FDMA), does not impose great overheads on the network (as is the case in TDMA) or does not assign each node a unique encoding scheme (as is the case in CDMA). However, CSMA-CA suffers from high collision probabilities and considerable packet delays which are primarily due to the limited design space of most of the existing CSMA-CA protocols. In particular, most of these protocols require all nodes to generate their back-off intervals using a uniform distribution with
M. Baz, PhD Thesis, University of York 2014
the same width and to double this width after each collision or channel assessment failure. Hence, the time required to service a packet increases multiplicatively based on the history of unsuccessful transmissions or failed channel assessments. Operating such a scheme over multihop WSNs makes the situation even worse, as in these networks, each node has distinct traffic characterises that differ substantially from others which in turn requires each node to contend to access the channel differently. Furthermore, in multihop networks the collision probability of each transmitter– receiver pair differs significantly than others as each transmitter and receiver nodes have different number of neighbours and these neighbours contend to access the channel according to different traffic loads. As a result using the same contention procedures by all nodes as the case with existing CSMA-CA scheme can be a source of performance degradation. Motivated by these limitations, this thesis proposes a novel MAC protocol that re-engineers the CSMA-CA scheme by replacing the uniform distribution with the Gamma distribution. The main aim of using the Gamma distribution is to provide a flexible means to generate the back-off intervals that can reflect the distinct contention characteristic of each node. The proposed CSMA-CA protocol also specifies an intelligent collision resolution mechanism that can identify the cause of the collision and remediate them instead of just doubling the back-off width blindly as the case in traditional CSMA-CA protocols. Interestingly, the proposed protocol exploits the principles of inferential statistics to allow a node to adjust the parameters of the Gamma distribution according to prediction for the status of the channel. The main advantage of the proposed protocol is that it reduces the end-to-end delay of packets as it allows a node to access an idle channel with high probability which in turn reduces the service time of packets and the associated contention energy. Moreover the proposed protocol increases the throughput of the network due to the intelligent collision resolution mechanism.
Reviewing the literature related to duty cycle management schemes demonstrates that most of these schemes either require nodes to be fitted with special equipment (as is the case with asynchronous or geographical cycle schemes) or impose overhead (as is the case with synchronous duty cycle schemes). As both of these approaches demand high energy consumption, the amount of energy saved by them is reduced significantly by the additional overhead or equipment required to maintain their operations correctly. Another main limitation of most duty cycle schemes is that they are devised as a part of a MAC or routing protocol which in turn makes their adaption to other configurations infeasible in some cases. Motivated by these limitations, this thesis proposes a novel duty cycle management scheme that can operate independently of routing or MAC protocols and that is applicable to a general multihop network without restrictions on the topology configurations, traffic patterns or routing policies. Basically, the proposed scheme enables a node to determine the time and durations of both sleeping and waking intervals individually using a nonparametric Bayesian inference technique in which each node infers the potential communications and adapts its duty cycle without need to use a control packet or special equipment. This approach is derived based on the metaphors between the wireless communication networks and chemical reaction networks. The main benefit of the proposed scheme is that it imposes no overheads on the network
M. Baz, PhD Thesis, University of York 2014
which in turn saves energy consumed in exchanging control packets and eliminates collisions between these control packets or between them and data packets. This in turn leads to further energy saving, reduction in the end-to-end delay of packets and magnified in throughput of network.
Finally, it can be seen from reviewing literature of routing protocols that the main challenge of developing an effective routing protocol for WSNs is that determination of the optimal routes requires fresh and accurate information to be gathered about all possible routes to the intended destination. However, obtaining such information imposes a great overhead which in turn can lessen the lifetime of nodes, reduce the throughput of network and increase the end-to-end delay of packets. Most of the routing approaches mitigate the effect of overheads either by reducing the size of routing information (as is the case with link state routing protocols) or by letting a node collect the routing information only when a route is required (as is the case with reactive routing protocols). However, such reduction increases the route acquisition latency problem. Motivated by these limitations, this thesis proposes a novel routing protocol that formulates routing as a solution for multi-objective optimisation problem in which each node attempts to route its packets over those paths that maintained the global optimisation for the entire network. The proposed protocol exploits one of the state-of-the-art optimisation algorithms, called the Harmony Search (HS) algorithm, to develop the routing protocol. The proposed protocol mimics the thinking strategy and logic reasoning of a musician ensemble during improvisation of the most pleasing harmony. Moreover, the proposed protocol is lightweight, scalable and highly adaptive as it allows each node to infer the routing metrics from the characteristics of its neighbour and utilises the principle of the spatial reasoning to guide a node to discover those areas of networks that presumably yield the optimal or near optimal paths. In additional, the routing protocol provides an error-correction mechanism that reduces the possibility of routing packets over suboptimal paths. The main advantage of the proposed protocol is that prolong the lifetime of nodes, reduce the end-to-end delay of packet and increase the throughput of network
3-10 Conclusion
This chapter has introduced MAC, routing, duty cycle schemes as well as their cross layer protocols with the aim to explore the fundamental logic of these protocols. Based on this exploration, some example protocols are reviewed in order to highlight their main contributions and limitations. This chapter has also highlighted the main motivation for the work presented in this thesis.
M. Baz, PhD Thesis, University of York 2014