A Weakly Secure Privacy Preserving Mechanism for Crowd Sensing

The popularity of smart sensing devices is greatly convenient for extensive collection of information. The information collected may include providers' private message, like location, personal identity, and so on. For protecting information providers' privacy, we design the information transfer mechanism WSNC (Weak Security Network Coding) integrated into routing policy. In this mechanism, we design the MWH (Maximizes Wi-Fi Hot), which improve reliability of data transmission. On the basis, we propose privacy protection algorithms based on WSNC which can provide privacy protection. The simulation shows that the privacy protection mechanism based on WSNC not only provides privacy protection also enhances the message delivery ratio. Introduction With the development of mobile communication and perceive technology, many innovative applications and services are emerging [1, 2]. They enlarge the dimensions and change the method that people sense the world. Also they open up the new areas of mobile internet, which is called MCS (Mobile Crowd Sensing) [3, 4]. The model of MCS provides a new way for people to participate sensing process while mobile terminals carry out sensing tasks. We transmit the data that we collected to sensing devices it encounter via Bluetooth technology. Then these sensing devices will send the data they stored to server via Wi-Fi when they move into the area where is the Wi-Fi coverage, which will avoid consuming 3G/4G data flow. The popularization of Wi-Fi provides an extremely good condition to transfer information efficiently. Based on this, we design MWH mobility model. We select the path which passes most Wi-Fi hot to the destination depending on threshold value of the mobility model. Those people represented by Tracey Ho [5, 6] proposed randomly linear network coding based network coding. When using randomly linear network coding, nodes can select coefficient randomly in the finite fields to code received or stored information and then forward it. Data transmission between sensing devices informs the Mobile Opportunistic Networks (MONs) [7]. We add WSNC into existing routing protocol [8] and propose the privacy protection mechanism in MONs. Once received information, the sensor node will encode it and then forward it to adjacent nodes. At the same time, the original data achieving the purpose of privacy protection. When receives enough information, server can decode the original information. Related Work Recently, many scholars have done a lot of works in MCS. In [9], authors proposed an efficient information diffusion protocol for sensor data collection via an opportunistic network and this protocol could minimize the information delivery delay. The privacy protection problem in the process of crowd sensing attracts a large researchers attentions. In the paper [10], the authors proposed TAPS to improve the quality of the collected sensor readings by optimizing the participants’ selection in proximity of a particular location of interest while still protecting their location privacy. There are some researchers focus on adding noise on the participants’ sensor reading to provide privacy, the authors [11] designed PESP and ALPS. 2nd International Conference on Advances in Mechanical Engineering and Industrial Informatics (AMEII 2016) © 2016. The authors Published by Atlantis Press 56 Mobility Model There exist three types of nodes in the network model: (1) sensor node V= {vi | 1 ≤ i ≤ n }; (2) Wi-Fi node W={wi | 1 ≤ i ≤ m}; (3) server node F={f}. Sensor node is responsible for sensing and forwarding data, sensor nodes transfer information with Bluetooth, sensor node will deliver the stored data to server node by Wi-Fi once meet Wi-Fi. The length of shortest path between two point A and B with Map based Dijkstra algorithm in the map denoted as D(A,B), the current location of people is Ls, the destination is Le, D0=D(Ls, Le). The total path length from Ls to Le denoted as L. If the path pass wi, we set P(wi) =1, otherwise P(wi)=0. The idea of MWH is that when LD0 ≤ β, maximize 1 ( ) i i P w δ = ∑ , β is a threshold. We use backtracking method to solve this problem, and this problem could turn solve the ordered subsets of W. We set (x1,...,xn) is an answer, xi means P(wi)=1. Then, we have the follow equation Eq. 1: L=D(Ls, 1 x w ) + 1 1 1 ( , ) i i n x x i D w w − + = ∑ + D( n x w ,Le). (1) When we choose (x1,......,xk-1), the qualification which we choose xk is xk is not equal from x1 to xk-1 and D(Ls, 1 x w ) + 1 1 1 ( , ) i i k x x i D w w − + = ∑ + D( n x w ,Le) D0 ≤ β. Weak Security Network Coding A. Weak Security Network Coding Algorithm The analysis of this section is based on the following information model: Definition 1 (information receiving model) Node u ∈ V, information m = ( 1 2 , , , n α α α L ) 1 2 ( , , , ) n x x x L arrive u, information m can be sensed by node u or relayed by other sensor node. X= 1 2 ( , , , ) T n x x x L , 1 2 ( , , , ) n α α α L is code vector, note as A0, m= A0X. The information packet node u stored which is the same generation as m is M, line i is the i-th information packet, denoted as Mi, A is M coefficient matrix, M = AX. Now connect A with A0 as a new matrix, denoted as A△=( A0, A). We design a cache strategy that node u will discard m when Rank(A△) ≠t+1. M will be divided into two set C and D, the i-th information in set C denoted as Ci, the number of messages in set C denoted as |C| and the number of messages in set D denoted as |D|. C stores encoded information of M, and uncoded information of M will be put in D. If M = ∅ , message m will be stored in u. When M≠∅ , two situations will be discussed: (1) Information m is sensed by node u. if C ≠∅ , put message m, D and Ci into new set H every time, then carry on random linear network coding with encode function Code(H), the number of elements in C is |C|, then get |C| new information by |C| times encoding. If C =∅ , D and m will be merged into H and get |D| messages by |D| times coding with encoding function Code(H). (2) Information m is relayed by other sensor nodes. If D ≠ ∅ , D and m will be merged into H and get |D| messages by coding |D| times with encoding function Code(H). If C ≠∅ , put m and Ci into new set H and then carry on random linear network coding with encoding function Code(H), the number of elements in C is |C|, then get |C| new information by |C| times coding. Finally, deleting the original information in M and saving the information encoded in node u. Coding function Code(H), a linear combination of information by random linear network coding in the finite field, all operations are completed in q  . If H= 1 2 1 1 1 , ,..., ) n n n l j j j j j j j j j g x g x g x = = = ∑ ∑ ∑ ( , when conduct random linear network coding, select l coefficients 1 2 , , , l k k k L from the finite field randomly, after encoding, get information packets Z= 1 1 ( ) l n i i j j i j k g x