Evaluation of SC-MAC

In this section, we compare our proposed schedule-based approach called SC-MAC (a SC-MAC protocol with Spatial Correlation consideration) with several schemes.

In the CSMA scheme, a CSMA-based MAC protocol without any spatial correla-tion consideracorrela-tion is adopted. In the CSMA-SSC (a CSMA-based protocol with Simple Spatial Correlation consideration) scheme, a CSMA-based MAC protocol with a simple report reduction scheme is adopted. This report reduction scheme works as that used in CC-MAC does. More precisely, when a node, say x, over-hears a packet whose reporter is y, x will judge whether the distance between itself and y is smaller than correlation radius or not. If the answer is affirma-tive, x will suspend its report. Otherwise, x will continue its report. Thus, the CSMA-SSC scheme can be viewed as the simplified version of CC-MAC. In the SCMAC-SCC-TSA scheme, the aforementioned report reduction scheme and our

proposed schedule-based approach will be adopted; however, the Traditional Slot Assignment strategy (i.e., a node needs to own a slot different from those used by all of its one-hop and two-hop neighbors) is used. In the SCMAC-TSA scheme, our proposed schedule-based approach with the traditional slot assignment strat-egy is adopted. In the SCMAC-SSC, our proposed schedule-based approach with the simple report reduction scheme is adopted. Finally, one should note that the RTS/CTS mechanism is not used in all schemes. In addition, the acknowledge-ment scheme is also disabled in the simulation.

Fig. 6.9 shows the results of the case where R_corr > R_tran (R_corr is 15 units, R_tran is 10 units, and α is set to be 0.5 for the SCMAC-TSA and the SCMAC schemes). To begin with, we focus on the CSMA and CSMA-SSC schemes. Al-though they have the best performance in terms of average delay, they have the worst performance in terms of coverage, especially when the event area becomes larger. The reason is high contention and collision that make the sink receive fewer reports than expected. This can be further verified by Fig. 6.9(b). We can see that the success rate is low when the CSMA and the CSMA-SSC schemes are adopted. Then, we focus on the SCMAC-SCC-TSA and SCMAC-SCC schemes.

In Fig. 6.9(a), we can see that both of them will transmit many packets in the network but they do not provide better coverage than the SCMAC scheme does.

The reason can be explained by Fig. 6.9(d). We can see that redundant reports are too much when these two schemes are applied. This means that the simple re-port reduction scheme does not perform well enough. In addition, because many reports have to be sent, these two schemes also do not perform well in terms of average delay. Finally, we focus on the SCMAC-TSA and SCMAC schemes. In fact, it is hard to compare these two schemes fairly. Although both of them set α to be 0.5, different slot assignment strategies may result in different performances.

However, it is not hard to see that the SCMAC scheme has better performance in terms of average delay (note that we can see that the SCMAC scheme transmits more packets than the SCMAC-TSA scheme does). This demonstrates that the proposed slot assignment strategy can reduce the report latency significantly (this

is because the value of MAXSLOT is reduced). Briefly, our proposed SCMAC scheme has the best performance. It provides high success rate, high coverage, low redundancy, reasonable average delay, and reasonable amount of packets.

Fig. 6.10 shows the results of the case where R_corr < R_tran (R_corr is 5 units, Rtran is 10 units, and α is set to be 1.0 for the SCMAC-TSA and the SCMAC schemes). In this case, we observe that the coverage will be low when α is set to be smaller than 1.0. Thus, α is set to be 1.0. We can see that the SCMAC scheme is still the best one. In addition, we can further note that the advantage of our proposed slot assignment strategy is revealed thoroughly in this experiment. In Fig. 6.10(c), we can see that our proposed slot assignment strategy can reduce the average delay. This also influences the coverage. In Fig. 6.10(d), we can see that the coverage will become lower when the event area becomes larger. One reason is buffer overflow (note that we assume that each sensor’s sending buffer is limited such that for a sensor, if there are too many packets to be sent simultaneously, some of packets will be discarded), because more report packets have to be sent when the event area becomes larger. Long delay will worsen the buffer overflow problem, because packets will be queued in a sensor for a long time. Thus, the performance of the SCMAC and SCMAC-SSC schemes is better than that of the SCMAC-SSC-TSA and SCMAC-TSA schemes.

To conclude, the advantages of our proposed SCMAC scheme can be summa-rized as follows:

• Our proposed medium access scheme can relieve the collision problem without the aid of RTS/CTS mechanism. This can be verified by high suc-cess rate.

• Our proposed report reduction scheme can reduce more redundant reports than the simple report reduction scheme does.

• Our proposed slot assignment strategy can shorten the average delay. This also relieves the buffer overflow problem.