Objectives

Mitik proposes to rely on non-intrusive passive measurements to infer the mobility of nodes and their potential interactions while on the move. The idea is to leverage the fact that mobile devices are likely to transmit Bluetooth and Wi-Fi packets even when users are not explicitly manipulating their devices.

We propose to measure and analyze such transmissions to infer the displacements of the nodes (see the architecture in Figure 1). On the other hand, passive measurement does not come for free (sniffers either capture a packet, or do not), achieving precise mobility characterization while respecting user privacy is challenging. To cope with this specificity, Mitik will adopt a multi-technique methodology involving:

  • optimization techniques for efficient placement of sniffers
  • anomaly detection approaches for filtering of noisy wireless measures
  • advanced synchronization strategies for the merging of measurement data from multiple sniffers
  • techniques for reconstruction of imprecise trajectories
  • contact inference from rough estimations of trajectories, and, at last, but not least
  • Mitik will be made fully respectful of user privacy
Passive data collection

Passive data collection

At its core, Mitik’s primary objective is the design of an entirely new methodology to help the community obtain real wireless contact traces that are non-intrusive, representative, and independent of third parties. The secondary outcome of Mitik would be the public release of:

  • the measurement tool designed for the easy contact gathering task
  • contact traces which are clean, processed, and privacy-preserving, i.e., protecting both the anonymity and the location privacy of the user and their spatiotemporal statistical analysis.
  • we plan to confront the new insights resulted from our analysis with established observations found in the literature, which may invite the research community to revisit opportunistic mobility modeling. We expect Mitik outcomes will support non-biased research on the modeling as well as on the leveraging of wireless contact patterns.

To achieve these objectives, the consortium will face many scientific and technological challenges:

  • Challenge 1 – Sniffers infrastructure design. Mitik’s measurement strategy relies on the physical deployment of passive sniffers over a geographical area. This will require answering the following questions: how many, where, and what type of sniffers have to be deployed? Such questions reveal the need for protocol and hardware specification as well as for optimized placement strategies. Those require taking into consideration factors such as the size of the monitored area, the type of traffic to be measured, and the measurement capabilities of sniffers, to cite a few.
  • Challenge 2 – Trace handling, sanitization, and merging. The spatiotemporal aspects of distributed passive measurement make the handling and merging of traces collected by each sniffer a challenging task. This requires advanced approaches for error handling, filtering, and synchronization, packet parsing, as well as merging of traces coming from different sniffers. Furthermore, we want to ensure that the privacy of the users is adequately protected under the GDPR guidelines, i.e., it should not be possible to retrieve the identity of the users from the traces we make available. In this regard, we will anonymize users’ identifiers as well as apply controlled noise to any users’ location. This should be done while minimizing the impact on the utility of the Mitik outcomes.
  • Challenge 3 – Trajectory reconstruction and contact inference. Two aspects make the computation of the exact trajectories of individuals a challenging task, namely the communication uncertainties brought by the wireless medium and the unfeasibility of measuring the precise geographical distances between devices and sniffers. Hence, instead of trajectories that are perfect lines, we obtain what we call a trajectory envelop (see the rose and green trajectories in Figure 2 ). The shapes of such envelopes depend on several parameters, ranging from the communication technology of nodes (e.g., Bluetooth’s physical coverage differentiates from Wi-Fi’s) to the density and positioning of sniffers. This makes the inference of contacts from such imprecise trajectories a difficult exercise.
Trajectory envelopes and plausible contact zone

Trajectory envelopes and plausible contact zone

Comments are closed.