Wireless communications are becoming ubiquitous. Wi-Fi and third-generation (3G) or fourth-generation (4G) connections are everywhere. For Wi-Fi, you need access points. For 3G or 4G, you need base stations for your cellular phones to connect. As the demand grows, so do the deployments of base stations. It is estimated that base stations account for 60 to 80 percent of the total network energy consumption. This paper proposes a distributed base station activation method for saving base station energies, and hence the operation costs. It uses a distributed algorithm based on Lagrangian dual decomposition, in which each base station computes its optimal activation probability through message exchange with its neighbors.

Besides saving base station energies, there are two other things to consider. The first is user satisfaction about the service. The second is whether the frequent on-off will hurt the base station electronics, thus increasing the maintenance cost. For evaluating the quality of a particular active set of base stations with respect to network users’ satisfaction, the paper models user satisfaction via the notion of utilities. A utility function is associated to each user, where the user utility is an increasing function of the amount of resources allocated to the user.

However, response time is more important to users. The frequent on-off of base stations will increase response time since rebooting a base station takes time. The authors should also consider it in the modeling. Another thought is to design a multiple-power-state base station, not just on-off state. Then, depending on the traffic, the system could have more energy-efficient system control without jeopardizing performance.

Finally, will frequent on-off reduce the lifetime of a base station? Also, what is the best size of an area for this algorithm to work? The simulation uses a square of 1,200 meters. Is it optimal?

This paper could be improved by including more useful and relevant citations. There is much more research available about base station power consumption than the authors cite.