DRIVEN: a Framework for Efficient Data Retrieval and Clustering in Vehicular Networks

A New Pattern Representation Method for Time-Series Data

“…Each segment is represented by a function which can be a linear regression, a line which is an approximation of the curve, or interpolation through the endpoint of each sequence or in other words, a line which approximates the curve with including the endpoints [15]. There are existing works in introducing the streaming version of this technique [16]. Duvignau et al [17] implemented a new Piecewise Linear Approximation (PLA) method in streaming paradigm with introducing a "singleton stream" contains only one value and using the number of points in each segment rather than storing the entire timestamps.…”

Time- and Computation-Efficient Data Localization at Vehicular Networks’ Edge

“…On the other hand, contacting too many vehicles might result in some of them wasting some of their computational power to inspect data that is not actually needed by the analyst. Thus, we need to require just enough positive answers (e.g., for statistical significance or for reducing the likelihood of identifying individuals in the data), but not too many (because of the time needed to collect all the data [20], [7], the induced computational load, and potential network stress).…”

“…The second dataset consists of CAN data and GPS traces from 20 hybrid cars internally collected by Volvo Car Corporation [20], [7] in the year 2015. After pre-processing, we generate 3462 trace files, each corresponding to a daily usage of one vehicle (cf.…”

“…Similarly to how Mobile Edge Computing [4] pushes parts of data analysis applications, previously run entirely in the cloud, towards mobile users, to achieve lower latency and bandwidth consumption, Vehicular Edge Computing [5] aims at better utilizing the cumulative computational power of Vehicular Networks while coping with high-mobility networks and the challenges stemming from their dynamic topologies and communications. When focusing on data analysis in the context of Vehicular Networks, a critical challenge is that of data gathering [6], [7] for subsequent analysis. While in the past companies could potentially afford the central gathering of all the data sensed by an entire fleet of vehicles (see [8], [9] for applications and services in VANETs), a modern vehicle can now generate several gigabytes of data per hour [1], making this approach infeasible in terms of infrastructure and costs -which could be alleviated by enforcing the collection of just enough data from relevant vehicles only.…”

“…While in the past companies could potentially afford the central gathering of all the data sensed by an entire fleet of vehicles (see [8], [9] for applications and services in VANETs), a modern vehicle can now generate several gigabytes of data per hour [1], making this approach infeasible in terms of infrastructure and costs -which could be alleviated by enforcing the collection of just enough data from relevant vehicles only. Several recent studies in the literature are focusing on how to avoid general central data gathering by transitioning to models in which the data selection processes, or even the analysis itself, are pushed towards the vehicles [10], [11], [12], for efficient and continuous filtering [13], preprocessing through online compression [14], [7], and conversion of raw data into information in Federated Learning [15]. However, these distributed analysis models often lack mechanisms that attempt to involve only valid vehicles into the analysis, to thus avoid unnecessary computational load and data transfers VOLUME X, 2021 or other hindrances to the analysis.…”