Privacy-Aware Incentivisation for Participatory Sensing

Abstract

Participatory sensing is a paradigm through which mobile device users (or participants) collect and share data about their environments. The data captured by participants is typically submitted to an intermediary (the service provider) who will build a service based upon this data. For a participatory sensing system to attract the data submissions it requires, its users often need to be incentivised. Such an incentivisation mechanism typically requires users to at least partially disclose their identity to be able to reward them, and to ensure that they are only rewarded for truthful submissions (called incentive compatibility). This, however, might deter privacy conscious users from participating. Therefore, an incentivisation mechanism needs to support anonymous and unlinkable data submission and untraceable and unlinkable rewarding while also ensuring data truthfulness (An incentivisation scheme is not in and of itself incentive compatible but should be able to facilitate incentive compatibility). Furthermore, as an environment can quickly and suddenly change (for example, an accident causing elevated pollution levels and a buildup of traffic), the value of a given data item to the service provider is likely to change significantly over time, and therefore an incentivisation scheme must be able to adapt the rewards it offers in real-time to match the environmental conditions and current participation rates, thereby optimising the consumption of the service provider's budget. There are numerous approaches in the state of the art in the areas of identity privacy and incentivisation for participatory sensing. For example, one approach proposes a digital currency to enable participants to make data submissions without disclosing their privacy, while another approach devises tokens to exchange data and rewards in a privacy-preserving fashion. However, while basic identity privacy may be preserved in these approaches, other forms of identity privacy such as behavioural habits and frequent trajectories can be inferred from the submitted data, with some proposed solutions actually increasing the threat of such inference attacks occurring. Furthermore, none of the approaches in the state of the art support adapting the reward to match the environmental conditions. This thesis presents Privacy-Aware Incentivisation (PAI), which is a decentralised peer-to-peer exchange to enable anonymous and unlinkable data submission, untraceable and unlinkable reward allocation and spending, and adaptive incentive-compatible reward computation. This is achieved through the modification and extension of the concept of decentralised trading for cryptocurrencies to make payments (i.e. rewards) sent to a recipient (i.e. the participant)untraceable. Furthermore, the use of the Diffie-Hellman Exchange Protocol is modified to enable participants to create their own untraceable reward currency in the form of tokens to which the service provider can then assign value. Finally,the Lyapunov Optimisation method is used to create an adaptive reward allocation model that optimises the consumption of the service provider's budget. The principal contributions of PAI are: 1. A platform for anonymous and unlinkable data submission and untraceable and unlinkable rewarding that is robust to inference attacks from semi-honest service providers and other potential attackers. 2. A privacy-aware adaptive incentive-compatible incentivisation scheme. PAI is evaluated by proof and by comparing the approach to the most relevant approaches in the state of the art. The privacy robustness of PAI is demonstrated by proofs showing that participants can make anonymous and unlinkable data submissions to the service provider and receive untraceable and unlinkable rewards in return. The incentive compatibility of PAI is also demonstrated by proofs showing that rewards will not be allocated for data submissions that are deemed to be non truthful with the privacy preserving character of the incentive compatibility approach also being proven. The adaptiveness and budget consumption of PAI's adaptive reward allocation method is compared with the most relevant approaches in the state of the art for reward computation using experiments carried out in a simulated participatory sensing environment. The results of these experiments are, in general, favorable with the reward allocation method adapting in a more timely fashion compared to similar approaches. Experiments are also conducted to compare the performance and computational complexity of PAI with the most relevant privacy preserving incentivisation methods proposed in the state of the art. The results of these experiments find that, in general, PAI's energy consumption is less than that of other privacy preserving incentivisation methods while its core algorithms require less resources.