Monitoring energy usage on device level: centralized solutions


Markus Weiss


March 23, 2010

Report due:

March 16, 2010 (First version)
April, 6 2010 (Camera ready version)


Traditional households electricity meters provide a measure of the total amount of electricty consumed within a given period of time. The information about energy consumption is then retrieved by the electricity provider and communicated to the consumer for billing purposes. Since only the total consumption over a typically large period of time (months) is known, neither the provider nor the consumer can discriminate the amount of electricty consumed by specific devices. Although most consumers typically know their electricty hogs by name, several recent studies showed that providing real-time information about energy usage of single devices may induce a significant reduction of the total electricty consumption in households. Providing information about electricty consumption on device-level can indeed rise the awareness of consumers and induce them to reduce their overall consumption in order to save both money and the environment. Furthermore, since the pricing of electricty is poised to become dynamic, the value of real-time information will rise , since its availability will allow to ascertain the actual cost of operating a specific device at a certain time of the day.

Technical solutions to monitor electricity consumtpion on device-level can be classified in three main categories: centralized, distributed, and hybrid approaches. Centralized solutions rely solely on the profile of the total electricity consumption and possibly available a-priori information to retrieve the consumption due to single devices. To this end, the total consumption must be measured at a fine-grained temporal scale (e.g., every few seconds). A centralized solution thus only requires the availability of a (centralized) electricty meter that can monitor the total energy consumption of a household in real-time. Such an approach, although easy to implement, may be inaccurate. Instead, a distributed solution that measures consumption at the single devices may provide much more accurate information, but this typically also requires installing additional hardware in the households one wants to monitor. Hybrid solutions combine the benefits (and costs) of both approaches.

In the context of this talk, we will first illustrate how and how much providing information about electricty consumption on device-level can influence consumer behavior. We will then provide detailed description of two specific centralized approaches and discuss their feasibility and limitations.

Recommended bibliography:

ETH ZurichDistributed Systems Group
Last updated August 23 2010 01:54:45 PM MET ss