The mobile communications industry has been very good at minimizing the energy consumption of the mobile devices, but only recently due to increasing energy prices and the pressure of reducing carbon dioxide emissions has forced the industry to start paying more attention to the overall energy efficiency of the wireless access networks. Unfortunately, the green revivals come just too late to influence the LTE standardization.
Nowadays base stations consume very high power even if the traffic volumes are low or the transmission queues are completely empty. Most energy saving could by using discontinuous transmission DTX and reception DRX. DTX and RTX can be done in two time scales. Slow DTX and RTX switches down entire cells when there is no traffic. Small cells are mostly needed for providing high capacity while macro cells are used mainly for coverage. Hence during periods with low traffic micro, pico and femto cells could be switched off. Fast dynamic DTX and RTX would switch the transmission and reception off whenever there is no traffic to transmit. Unfortunately, dynamic DTX and RTX is cannot be implemented in the current networks without changing the way control information is broadcasted by the base stations making the approach non-backwards compatible with the existing terminals. If the industry would have been aware of the need of dynamic DTX amd RTX couple of years ago it would not have been difficult to include such features into the LTE standard. Now someone would need to come up with a clever hack that allows such feature to be used or we need to accept that some early LTE terminals will have performance problems when communicating with the evolved green LTE base stations…Or we need to keep wasting energy for the sake of compatibility.
Durign busy hour femto cells offer significant potential for energy saving. Macro cell traffic can be offloaded to femtos which can transmit the data more energy efficiently. At the same time the offered traffic to macro cells decrease allowing the operators to build the network with smaller number of sites. All traffic that is offloaded from the operators network help in reducing the base station energy consumption and thus OPEX. However, the overall energy saving depends on the fraction of traffic carried by femtos and their energy saving features. In a recent case study of WCDMA networks, we found that if the femtocells do not have efficient power save features off loading less than 20% of the overall traffic to them actually increases the overall energy consumption. In the worst case, introduction of femtos could almost double the overall energy consumption in the network! However, if the femtos are equipped with dynamic DTX and RTX features they would be helpful in reducing the energy consumption even is just 5% of the macro cell traffic is offloaded to them. The main message here is that femto cells need efficient power save mechanisms, otherwise they can have negative impact on the overall energy consumption of the networks.
You are right – massive deployment of femtocells without good idle power handling, could be disaster (probably is using WCDMA or LTE as is in femtos is a really bad idea). On the other hand femtos are needed for expanding peak-hour capacity in indoor cases. This means one could look for system level (slow time-constant) solutions, rather than modifying the phy/mac-layer. In many cases when coverage is not the problem you could simply and crudely switch the femtos off when traffic is low, saving 100% of their energy.
Even though 3GPP did not focus on energy savings in the initial release, we at Ericsson always have a low energy footprint in our minds. Rest assured that future releases of our LTE RAN products will get more and more energy efficient!
@Andreas: although Ericssons intentions are noble ;) , it not obvious to me that more and more energy efficient (i.e. an evolutionary improvement of the current systems as we see now ) we be enough. We are looking for several orders of magnitude increase in capacity demand and more “revolutionary” solutions as Riku discusses or other different system concepts may be necessary to take the next steps.