A practical sleep coordination and management scheme with duty cycle control for energy sustainable IEEE 802.11s wireless mesh networks

Abstract

We consider the energy sustainable operation of solar powered IEEE 802.11s wireless mesh networks. Our main contribution is the development of a simple and novel sleep scheduling scheme that is local and distributed and provides contiguous sleep intervals that can be used for putting both radio interface cards and the main-board into deep sleep mode. We show this provides substantial energy savings as main-board power consumption comprises a significant portion of total node power. Unlike many sleep coordination schemes developed for Wireless Sensor Networks, our approach is suitable for Wireless Mesh Networks having much larger traffic demand and non-tree-like routing pattern. In addition, we propose a local duty-cycle control scheme, which regulates node awake time and naturally limits the amount of elastic traffic that moves along energy limited nodes. This is coupled with an implicit admission control scheme, which limits the number of non-elastic flows admitted to the network. More importantly, our scheme does not modify the IEEE 802.11 MAC and does not require information of the traffic demand nor input energy pattern. We have evaluated the performance of our approach using NS3 simulations by considering its traffic volume, lifetime and numerous other parameters and have also compared it to both perfect scheduling and default IEEE 802.11s behavior. Our results are also backed by evaluating numerous randomly generated topologies. A detailed discussion of the effect of topological aspects of the network on its sustainability characteristics is also provided.

Appendix: Example of required battery calculation

Recall that, $B^{req}$ is defined as the nominal minimum battery charge that should exist at the beginning of a given sustainability epoch. It can be determined based on any policy desired and according to any method, depending on the network design and operation objective. We treat $B^{req}$ as a requirement that should be met for the sustainability of the WMN, and one which is set outside the framework presented by our sleep coordination approach. Hence, the determination of $B^{req}$ is out of the scope of our work. For the sake of illustration, however, we include one such policy and calculation method.

Fig. 34

BM + critical connections and input energy profile

Fig. 35

Total input energy, total required energy for BM and critical connections and minimum required battery for energy sustainability

The example policy is to maintain $B^{req}$ enough for management and control traffic as well as some amount of critical traffic determined by the network administrator. Hence we have a minimum traffic profile which is translated into the amount of energy required to forward it at any MAP, along with an input energy profile which is based on historical irradiation data, all shown in Fig. 34. This information is then used according to the methodology presented in Eq. (21) to derive $B^{req}$. Figure 35 shows the minimum battery requirement for serving critical and control traffic until next daylight, which is calculated as follows. Let $P_{\min}(t)$ be the minimum power required by an MAP for its critical and control traffic. Here it is not assumed that an MAP is always awake. The power requirement of an MAP is calculated based on the amount of traffic it forwards, which translates into a certain activity time, hence energy consumption. To do this, the average power consumption in active mode is used. Take T as the end of current sustainability epoch and $T_{ND}$ as start time of the next daylight. If $\varGamma (t)$ is the harvested power, then the minimum required battery at the end of T is

$$B^{req}=\left[ \int _{T}^{T_{ND}}P_{\min}(t) dt-\int _{T}^{T_{ND}}\varGamma (t) dt \right] ^+ ,$$

(21)

Where $[x]^+$ is equal to x if $x\ge 0$ and 0 if $x\le 0$. Maintaining this $B^{req}$ will then ensure that the critical and control traffic is served until next daylight period.