An Identification Method of IR Signals to Collect Control Logs of Home Appliances

In order to realize a system that provides services according to the resident preferences, it is necessary to collect the daily life data of residents for a long term. Especially, the home appliances are closely related to the daily life of people. We can easily collect the control logs of the information home appliances that can be controlled through network. However, types of such appliances are not various and legacy home appliances are mixed in the real home environment. Most home appliances can be controlled by infrared (IR) remote signals. If we can identify the IR signals, we can collect the control logs of the heterogeneous appliances. Although it is difficult to identify the IR signals due to some reasons such as errors of signal, reception environments, etc. In this paper, we propose an identification method for IR signals with different length and format to achieve a system collecting home appliance control logs. Our method identifies the type of home appliance and the type of command by errors between IR signal from remote controller and each signal stored in database. We conducted 10-fold cross-validation for total 140 commands from 14 home appliances. As a result, our method achieved to identify 95.5% accuracy for the type of home appliance and 92% accuracy for the type of command.