Identifying line-of-sight (LOS) and non-LOS (NLOS) channel conditions canimprove the performance of many wireless applications, such as signalstrength-based localization algorithms. For this purpose, channel stateinformation (CSI) obtained by commodity IEEE 802.11n devices can be used,because it contains information about channel impulse response (CIR). However,because of the limited sampling rate of the devices, a high-resolution CIR isnot available, and it is difficult to detect the existence of an LOS path froma single CSI measurement, but it can be inferred from the variation pattern ofCSI over time. To this end, we propose a recurrent neural network (RNN) model,which takes a series of CSI to identify the corresponding channel condition. Wecollect numerous measurement data under an indoor office environment, train theproposed RNN model, and compare the performance with those of existing schemesthat use handcrafted features. The proposed method efficiently learns anon-linear relationship between input and output, and thus, yields highaccuracy even for data obtained in a very short period.
展开▼