Privacy-Aware Data Aggregation in Mobile Sensing with Fault Tolerance

Over the years capability of the mobile sensing devices like smart phones are increased in terms of capturing and sharing the information. This information can be useful if analyzed as an aggregated value or values. Classic examples are like traffic trend in particular area, medical information in particular locality etc. When such information is shared to an aggregator, user’s identity should be compromised and only information needs to be shared. To achieve this user’s identity Li, Cao and Porte [1] has suggested efficient protocol to obtain sum aggregate value using homomorphism encryption and novel key management. Unfortunately, the use of cryptography in the projected protocol scheme introduced the all-or-nothing decryption model. Therefore, the aggregator learns nothing if a single user fails. In some of the critical applications, data has to be collected from all users to get right results. Any leakage in data will lead to incorrect aggression and sub-sequent calculations. Fault tolerance has to be handled to get right results. T.-H.H. Chan, E. Shi, and D. Song [4] have proposed a technique to handle fault tolerance. In proposed technique, a binary interval tree over n users, and allow the aggregator to estimate the sum of contiguous intervals of users as represented by nodes in the interval tree. The binary-tree technique allows handling user failures joins and leaves, with a small logarithmic (or polylog) cost in terms of communication and estimation error. Keywords—Secure computing, Mobile Sensing, Privacy, Data Aggregation, Fault Tolerance.