As we described in Section 5.2.4, two update schemes, proactive and on-demand, are proposed for the case where new nodes are deployed. Compared with on-demand update scheme, proac- tive update scheme consumes much more energy because the new deployed nodes actively send the preamble packets. As a result, we only use the on-demand update scheme in this sim- ulation. 100 nodes are deployed and we deploy a certain percentage of new nodes when the network is in run-time phase. The percentage is changed from 5% to 50%.
It is shown in Figure 5.16(a) that the average delivery ratio is decreasing as the percentage of new nodes is increasing. The average delivery ratio is about 0.9 when there are no new nodes while the average delivery ratio is about 0.7 when there are 50% of new node (i.e. 50 new deployed nodes). And we can see from Figure 5.16(b) and Figure 5.16(c) that the average end- to-end delay and the average transmitted packets are not changed so much when we increase the percentage of new deployed nodes. Those simulation results show that RF is a highly robust
0 0.2 0.4 0.6 0.8 1 0 5 10 20 25 30 40 50
Average delivery ratio
Percentage of new nodes (%) RF
(a) Average delivery ratio
0 2 4 6 8 10 0 5 10 20 25 30 40 50
Average end-to-end delay (s)
Percentage of new nodes (%) RF
(b) Average end-to-end delay
0 2000 4000 6000 8000 10000 12000 14000 0 5 10 20 25 30 40 50
Average transmitted packets
Percentage of new nodes (%) RF
(c) Average transmitted packet
Figure 5.16: Impacts of the number of new added nodes protocol.
5.4
Summary
In this chapter, we have presented a robust forwarding protocol (RF) for ultra-low duty-cycle WSNs. Because the EDR metric used in WSEDR is not easy to update, RF adopts RSSI as the key metric for forwarding. The main advantages of RF are that the sender gives a higher priority to the candidate that has a lower COST and every node can dynamically update their COST information. We simulated RF with different settings and compared RF with SOFA. From the simulation results, we showed that RF has a higher average delivery ratio than SOFA, and the average end-to-end delay is a little higher, but the average number of transmitted packets are lower. We also compared RF with RTCF. The simulation results show that the performance of RTCF is better than that of RF when the network has no or less dead nodes. The performance of RF is better than RTCF when the network has many dead nodes.
Chapter 6
Conclusion and future work
6.1
Conclusion
Sensor nodes in WSNs are battery-powered, thus communication protocols have to be energy efficient in order to prolong the lifetime. Duty-cycle is one of the most energy efficient solutions for WSNs, but it induces other problems, in particular, longer delay. Some applications have real-time constraints, which means a packet must be sent to the sink node before a deadline. Moreover, due to the low-power transmission, wireless links among sensor nodes are highly unreliable. Furthermore, the topology in WSNs may be changed because of the dead sensor nodes or new added sensor nodes. This characteristic requires that the protocols for WSNs should be robust enough. In this thesis, considering the aforementioned problems, we have proposed some reliable and real-time constrained protocols for WSNs.
First, we reviewed the main existing protocols in WSNs in Chapter 2, including MAC, routing and cross-layer forwarding protocols. As shown in some experimental results, wire- less links among low-power sensor nodes are highly unreliable, which has significant negative effects on the performances of routing protocols. Besides the impacts of unreliable links, the impacts of duty-cycle on the performances of routing protocols can not be neglected. Thus cross-layer protocols can have better performances in those cases. In this chapter, we under- lined that routing over virtual coordinates can have very good performances.
The goal of Chapter 3 was to know the relationships between the three key points in order to design reliable and real-time constrained protocols, namely virtual coordinate, unreliable links and duty-cycle, as well as to get some guidelines. In Chapter 3, we first analyzed the impacts of duty-cycle MAC protocols on the performances of routing protocols. From the analyzed results, we showed that cross-layer based duty-cycle MAC protocols can significantly reduce the average end-to-end delay. In addition to the impacts of duty-cycle, we also studied the impacts of unreliable links on virtual coordinate based routing protocols. It was shown that it is better to take unreliable links into consideration when constructing the virtual coordinate. Then, we proposed three simple and efficient update schemes for constructing two dimensional
virtual coordinates. The performances of the routing protocol based on the virtual coordinate constructed by the proposed schemes were shown by large-scale simulations.
Based on the study results in Chapter 3, we proposed two versions of a cross-layer forward- ing protocols in Chapter 4, i.e. synchronous version and asynchronous version. We considered the impacts of duty-cycle and unreliable links simultaneously. In synchronous forwarding pro- tocol, we showed that different wakeup slots have significant impacts on the expected delivery ratio, and proposed a wakeup scheduling algorithm. The scheduling algorithm simply sched- ules the wakeup slots of every node according to their own hop count and expected delivery ratio. The performances of the proposed synchronous forwarding protocol were evaluated by large-scale simulations. The simulation results are in line with the mathematical ones. We then proposed an asynchronous version protocol. In asynchronous forwarding protocol, because ev- ery node is not synchronized to each other, preamble sampling technique was adopted to wake up the potential candidates. And the asynchronous forwarding protocol can dynamically adjust the number of potential candidates to increase the delivery ratio by considering the remaining time before the deadline. The performances of the proposed asynchronous forwarding protocol were shown by a large number of simulations.
The two cross-layer forwarding protocols can have good performances in the scenario where the topology is relatively stable. However, dynamicity is an important characteristic in some types of WSNs, for example, link quality is changing very fast, some nodes may die due to energy exhaustion and some new nodes may be added to the network. Because of the topology changes, it is necessary to design robust protocols for those WSNs. In Chapter 5, a robust forwarding (RF) protocol was proposed. RF can automatically update the network when the network topology is changing. The forwarding metric of RF is based on RSSI/LQI. The performances of RF were compared to SOFA and RTCF because, to the best of our knowledge, SOFA is the only forwarding protocol which takes the low duty-cycle into consideration. Sim- ulation results showed that RF can have much better performances when the network topology changes.