An enhanced feature detection method for DTV signals

This paper proposes a robust sensing method for detection of digital TV (DTV) signals. The proposed method exploits the newly discovered features of sliding correlation of DVB-T with its pilot signal for detection: some correlation peaks are observed only having slight difference of amplitudes and, distance between their positions in the sequence of sliding correlation outputs keep constant for different DVB-T timings; moreover, the feature applies for both positive correlation peaks and negative correlation peaks. The method then aggregates the peaks based on their positions and use the aggregation as test statistic; in particular, the aggregations are conducted for positive peaks and negative peaks separately firstly, two aggregation results are then further aggregated for their original values by subtraction. Analysis results show that the proposed method outperforms the conventional one with a slightly increased computational complexity. Computer simulation results further verify that for 2K mode DVB-T, the proposed method reduces sampling time by about 25% for a target sensing performance while increasing computational complexity by around 0.0001%. Hardware test performance further verifies that the proposed method is able to accurately detect ISDB-T at low SNR of -14.5dB, using samples of 8 OFDM symbol durations.