We present a survey of smartphone-based Transport Mode Detection (TMD). We categorize TMD solutions into local and remote; the first ones are addressed in this article. A local approach performs the following steps in the smartphone (and not in some faraway cloud servers): 1) data collection or sensing, 2) preprocessing, 3) feature extraction, and 4) classification (with a previous training phase). A local TMD approach outperforms a remote approach due to less delay, improved privacy, no need for Internet connection, better or equal accuracy and smaller data size. Therefore, we present local TMD solutions taking into account the above mentioned four steps and analyze them according to the most relevant requirements: accuracy, delay, resources consumption and generalization. To achieve the highest accuracy (100%), studies used a different combination of sensors, features and Machine Learning (ML) algorithms. The results suggest that accelerometer and GPS (Global Position System) are the most useful sensors for data collection. Discriminative ML algorithms, such as random forest, outperform the other algorithms for classification. Some solutions improved the delay of the proposed system by using a small window size and a local approach. A few studies could improve battery usage of their system by utilizing low battery-consuming sensors (e.g., accelerometer) and low sampling rate (e.g., 10Hz). CPU usage is primarily dependent on data collection, while memory usage is related to the features and complexity of the ML algorithm. Finally, the generalization requirement is met in studies that consider user, location and position independency into account.

The kNN machine learning method is widely used as a classifier in Human Activity Recognition (HAR) systems. Although the kNN algorithm works similarly both online and in offline mode, the use of all training instances is much more critical online than offline due to time and memory restrictions in the online mode. Some methods propose decreasing the high computational costs of kNN by focusing, e.g., on approximate kNN solutions such as the ones relying on Locality-Sensitive Hashing (LSH). However, embedded kNN implementations also need to address the target device’s memory constraints, especially as the use of online classification needs to cope with those constraints to be practical. This paper discusses online approaches to reduce the number of training instances stored in the kNN search space. To address practical implementations of HAR systems using kNN, this paper presents simple, energy/computationally efficient, and real-time feasible schemes to maintain at runtime a maximum number of training instances stored by kNN. The proposed schemes include policies for substituting the training instances, maintaining the search space to a maximum size. Experiments in the context of HAR datasets show the efficiency of our best schemes.

No abstract

No abstract