Detecting Anomalous Noise Events on Low-Capacity Acoustic Sensor in Dynamic Road Traffic Noise Mapping

Abstract

Abstract:One of the main aspects affecting the life of people living in urban and suburban areas is their continued exposure to high road traffic noise (RTN) levels, traditionally measured by specialists working on the field. Nowadays, the deployment of Wireless Acoustic Sensor Networks (WASN) has allowed to automate noise mapping in Smart Cities. In order to obtain a reliable picture of the RTN levels affecting citizens, those anomalous noise events (ANE) unrelated to road traffic should be removed from the noise map computation. For this purpose, an Anomalous Noise Event Detector (ANED) designed to differentiate in real-time between RTN and ANE should be developed to run on the low-cost acoustic sensors of the WASN. In this work, the viability of implementing the ANED algorithm to run on low-capacity (LowCap) µ controller-based acoustic sensors developed within the DYNAMAP project is presented, after being designed and implemented for the high-capacity sensors. The algorithm is based on the comparison between RTN and ANE spectral differences using real-life acoustic data from both suburban and urban scenarios. The results show significant spectral differences between RTN and ANE classes in both environments, after being parametrized using Gammatone Cepstral Coefficients. However, further research should be conducted to determine the most discriminant subbands, which should be taken into account for the implementation of the ANED LowCap version.

Optimized Sensors Network and Dynamical Maps for Monitoring Traffic Noise in a Large Urban Zone

“…For this reason, trying to recognize environmental sounds by just adapting well-established approaches to music or speech recognition is a suboptimal way to proceed. On the contrary, it becomes necessary to devise specific recognition strategies to tackle this kind of problem [71][72][73][74][75][76].…”

Detection of Anomalous Noise Events on Low-Capacity Acoustic Nodes for Dynamic Road Traffic Noise Mapping within an Hybrid WASN

“…In order to implement the algorithm in an affordable Lo-Cap platform with a C core, a change of paradigm on the AED is mandatory, both in terms of feature extraction and classification, to assume the severe decrease of the computational load of the Lo-Cap platform with respect to the Hi-Cap counterpart. As studied in [ 22 ], the distribution of the energy per subband show signs of separability depending on the frequency between ANE and RTN, and the proposal presented in this piece of research deepens the analysis in this direction.…”

“…The study of the spectral differences between the ANE and the RTN [ 22 ] allows us to conclude that, for each acoustic scenario, the separability of the two types of signal is possible, especially if the potential algorithm focuses on specific frequency ranges, as observed in Section 4.2 . The two evaluated acoustic scenarios, urban and suburban, do not present the same probability of error results in terms of separation of ANE and RTN and its associated MFS.…”

“…Once the error probability function has been obtained for a given acoustic dataset (see Equation ( 2 )), its evaluation can be used to define a suitable frequency range use in the classification in terms of ANE and RTN discrimination capabilities [ 22 ] (see the last module in Figure 3 ). This selection of frequency range is defined following the next simple rule: including those MFS subbands for which the error probability function is less or equal than a certain probability of error threshold.…”

“…The ANED Lo-Cap algorithm bases on the same principle of the original ANED version to discriminate between ANE and RTN to obtain consistent results across the nodes of the hybrid WASN, but adapted to fit the computational capacity of a Lo-Cap sensor. For this purpose, the approach follows a threshold-based binary classification scheme that considers the spectral energy differences between ANE and RTN [ 22 ] for the most discriminatory Mel-based Frequency Subbands (MFS). The experiments are conducted considering 9h and 8 min of real-life acoustic data from the two pilot areas (urban and suburban) of the DYNAMAP project.…”