Broad Learning System-Based Detector for OFDM With Index Modulation

Abstract

This letter presents a broad learning system (BLS) based orthogonal frequency division multiplexing with index modulation (OFDM-IM) detector called BLS-IM, which achieves bit-error-rate performance improvement with less training time. In BLS-IM, in order to help the network detect the index of the active subcarrier, the received data needs to be preprocessed into polar coordinates first. The processed data is then fed into the BLS network. In particular, the BLS is built in the form of a flat network where the initial signal bits are transferred among the feature nodes. At the same time, the structure is extended in a broad sense in terms of enhanced nodes. The proposed BLS-IM detector has an advantage in training time because it avoids the time-consuming training process caused by a large number of connection parameters in the layers. Simulation results demonstrate that BLS-IM is capable of achieving better detection performance with less training time than the existing algorithms.