17-10-2016, 03:04 PM
PRIVACY PRESERVING PERSONALISED AD-DISSEMINATION BASED ON INTEREST AGGREGATION AND PIGGY BACKING
1459446886-PRIVACYPRESERVINGPERSONALISEDAD1.docx (Size: 53.96 KB / Downloads: 5)
ABSTRACT:
Mobile devices get more involved as media delivery platforms; the worth of advertising on these devices becomes significant. Many systems have been developed to make use of this opportunity. In existing system the peers among the group can easily access the request of the particular peer which is not a valid process on ad server. This leads to the lack of privacy in the system. In order to overcome the existing process a new model is proposed to enhance a security process of previous system and to reduce the communication cost. The proposed system has three types of roles like service provider, content provider and mobile peers. Service provider provides the advertisement to Ad server. Ad server distributes the advertisement to the content provider. Mobile peers (user) install the third party application. The peer group formation starts when a peer broad cast an ad announcement. In this process three different algorithms are used to overcome the problem of existing system. Signature key is generated using HMAC algorithm, the data transmission between the peers and ad server is done securely using BASE 64 algorithm and the energy efficiency in the primary peer is increased by using RSA algorithm. A new concept of re encryption process is introduced using the intermediate peer among the group of peers to enhance the security so that the peer id is hidden from the ad server and primary peer to avoid the unauthorized process.
INTRODUCTION:
Mobile advertising is a rapidly developing sector which provides brands, agencies and marketers the opportunity to connect with consumers. Users spend significant time browsing the different multimedia and gamming and get exposed to ads. The Customized advertisement matches with user preferences with product to achieve better customer satisfaction. These devices now come with Wi-Fi and 3G, meaning they can be reached virtually everywhere. Add to this GPS capability and computing user preferences, and a new level of targeted advertising can be attained. We propose a system for Mobile advertising relies on content providers like applications and WebPages to deliver ads to users. Service providers register ads to an ad server, which delivers them to users through content providers who usually subscribe to host ads for profit making. When a user accesses an application subscribed to an ad-server, the application requests an ad from the server with the user location and id. The server then checks based on the id the interests of the user through an online profile, and delivers targeted ads that refer to service providers in the vicinity of the user which are relevant to his interests.
For example, a user in downtown San Francisco interested in pizza will get an ad for pizzerias within that location. After the user clicks the delivered ad, a click Report is sent to the ad-server for billing purposes.
The present advertising model relies in the following classes of threats:
• Direct advertisement request in ad-server leads to lack of privacy.
• Expired ads of the user get received during shuffling process leads to cause of multiple participation.
• Algorithm used for encrypt and decrypt the message is not satisfactory.
• Overlapping of data occurs during multipleparticipation of expired ads.
BACKGROUND AND RELATED WORKS
To frame the problem, we describe how mobile advertising currently works and how the present scenario in advertisements leads to the privacy and security threats from malicious advertising and vulnerable advertising networks.
Related Works
Concurrent with this work, several other researchers have explained about mobile advertisement services which have explained as, [1]A system for delivering context, location, time, and preference-aware advertisements to mobiles. The main adversary in our model is the server distributing the ads, which is trying to identify users and track them, and to a lesser extent, other peers in the wireless network.
Here, Direct advertisement request in ad-server leads to lack of privacy. Algorithm used for encrypt and decrypt the message is not satisfactory. A Distributed mechanism for users to augment their profile in a way that confuses the user-item connection to an un-trusted server, with minimum loss on the accuracy of the recommender system. By using the method called Netflix prize dataset. provides a tool to separate the privilege given to advertisers in android from application requesting ads. Based on the notion of applications are granted the privilege of accessing the user’s preferences. Rapid expansion of wireless technologies has provided a platform to support intelligent systems in the domain of mobile marketing.
Personalized and context-aware advertisements to fulfill customer needs.Mobile Ad would perform a range of data mining tasks in order to maintain an interest profile on the user's phone, and use the infrastructure network to download and display relevant ads.[8] The system was designed to constantly deliver advertisements and information to wandering customers according to their location and previous visits. It is based on mobile advertising in a mall based on a hybrid system using a Bluetooth system. [10] Operates by grouping users into a large and geographically diverse group (crowd) that collectively issues requests on behalf of its members. It uses degrees of anonymity as an important tool for describing and proving anonymity properties
PROPOSED SYSTEM:
The proposed system is for users to aggregate user’s interests when requesting advertisements to hide user identities from the ad server. We developed three roles: Service Provider, Content Provider, and Mobile Peers. Service provider provides the advertisement to Ad-Server. Ad-server distributes the advertisements to the Content Provider. Mobile peers (user) install third party application. The peer group formation starts when a peer broadcasts an ad announcement. Peers who hear the message and need ads will reply with an acknowledgement and join the group. Some peers cannot hear the announcement, but can still hear the broadcast of peers that have joined the group. After choosing the primary peer, all participants in the group generate interests and encrypt these interests along with billing reports, which capture their clicks on previous ads, using the primary peer’s public key. With this process, peers hide their data from each other. Next, each peer randomly chooses another peer in the group and encrypts the encrypted message with his public key, before broadcasting it. With this mechanism, only that particular peer will be able to decrypt this message before transmitting it to the primary peer. As the primary peer receives these packets, it decrypts them using its private key, and aggregates them to be sent to the server. When the ad server receives the interests, it replies with ads to the primary peer, who will then broadcast them to the group.
The challenges to be focused by adding the following significant contributions:
• Enhanced security by introducing an intermediate peer among the group.
• Standard algorithm is used for encryption and decryption process for security purpose.
• An aggregation scheme and a piggyback method that protects the system from multiple participation of expired ads
ARCHITECTURE DESIGN ELABORATION:
The proposed system is for users to aggregate user’s interests when requesting advertisements to hide user identities from the ad server. We developed three roles: Service Provider, Content Provider, and Mobile Peers. Service provider provides the advertisement to Ad-Server. Ad-server distributes the advertisements to the Content Provider. Mobile peers (user) install third party application. As in fig1.1, The peer (mobile users) group formation starts when a peer broadcasts an ad announcement. Peers who hear the message and need ads will reply with an acknowledgement and join the group. After choosing the primary peer, all participants in the group generate interests and encrypt these interests along with billing reports, which capture their clicks on previous ads, using the primary peer’s public key. With this process, peers hide their data from each other. Next, each peer randomly chooses another peer which is called an intermediate peer in the group, it is chosen by referring higher priority.
The intermediate peer will encrypts the encrypted message with his public key, before broadcasting it. With this mechanism, only that particular peer will be able to decrypt this message before transmitting it to the primary peer. As the primary peer receives these packets, it decrypts them using its private key, and aggregates them to be sent to the server. When the ad server receives the interests, it replies with ads to the primary peer, who will then broadcast them to the group.
PROPOSED DESIGN MODEL
The main aim of this project is to provide users with personalized advertisements without affecting privacy from Ad-server. To provide benefits to the mobile users as well as Content Providers for viewing and disseminating advertisements respectively. Reduce the communication cost by piggybacking. To achieve this, few modules have been developed
A. POST ADVERTISEMENT
In this module, Service Provider and Content Provider have to register their details with the ad-server. After successful registration, details are stored in database. Service Provider login with their credentials and then post an advertisement to Ad-server with image, tags and benefits per clicks (both to content provider and user). Ad-server view the advertisements posted by the service provider and allocate the content provider.
B. PEERS FORMATION IN NETWORK
In this module, Peers are created based on coverage. Authority will generate public keys and private key for all peers using RSA algorithm. Public keys are distributed to all peers within coverage. The group formation starts when a peer broadcasts an ad announcement. Peers who receive the message and need ads will reply with an acknowledgement and join the group. Peer who one is acknowledged first then we selects that peer as primary peer.
ALGORITHM FOR PEERS FORMATION IN NETWORK
Step1: Choose two distinct prime numbers p and q.
Step2: Find n such that n = p*q.
(n will be used as the modulus for both the public and private keys)
Step3: Find the totient of n, Φ (n) = (p-1)(q-1).
Step4: Choose an e such that 1 < e < ϕ(n), and such that e and ϕ(n) share no divisors other than 1 (e and ϕ(n) are relatively prime).
(e is kept as the public key exponent)
Step5: Determine d (using modular arithmetic) which satisfies the congruence relation de = 1 (mod ϕ(n)).
Step6: Encryption c = me (mod n).
Step7: Decryption m = cd (mod n).
C. REQUEST AGGREGATION ON PRIMARY PEER:
Peer sends the advertisement request to server through primary peer and random choosing peer. Peer who is selected as a random peer will encrypt the advertisement using public key and forward to primary peer, then primary peer verifies the signature and then re-encrypts the advertisement. This re-encryption ensures the protection of data privacy and user privacy. Finally, after the primary peer receives all requests, it aggregates them and sends them to the ad server, and then waits for a reply. The ad server process the requests from the primary peer by finding the ads with metadata offering the best match to the tags contained in the message and replies back with the corresponding ads to the primary peer.
ALGORITHM FOR REQUEST AGGREGATION ON PRIMARY PEER
Step 1: If the length of K = B, set K0 = K. Go to step 4.
Step 2: If the length of K > B, hash K to obtain an L byte string: K = H(K).
Step: 3 If the length of K < B, append zeros to the end of K to create a B-byte string K0 (e.g., if K is 20 bytes in length and B = 64, then K will be appended with 44 zero bytes 0x00).
Step 4: Exclusive-Or K0 with ipad to produce a B-byte string: K0 Å ipad.
Step 5: Append the stream of data 'text' to the string resulting from step 4: (K0 Å ipad) |text.
Step 6: Apply H to the stream generated in step 5: H ((K0 Å ipad) | text).
Step 7: Exclusive-Or K0 with opad: K0 Å opad.
Step 8: Append the result from step 6 to step 7: (K0 Å opad) | H ((K0 Å ipad) | text).
Step 9: Apply H to the result from step 8: H ((K0 Å opad )| H((K0 Å ipad) | text)).
Step 10: Select the leftmost t bytes of the result of step 9 as the MAC
D. BILLING PROCESS AND PIGGYBACKING
Primary peer broadcast the advertisement to the peers within the coverage; only the requested mobile peers will receive the advertisement. Sybil attack could occur if a certain peer generates large amounts of “fake” click reports to charge service providers more. To rectify the Sybil attack if the peer generates a large amount of click reports ad-server will considered it as a one click. Piggybacking literally refers to carrying someone on one's back or shoulders. It may also refer to: Piggyback (transportation), something that is riding on the back of something else. Piggybacking (security), when an authorized person allows (intentionally or unintentionally) others to pass through a secure door. The ad server should be able to reliably bill service providers for the offered advertising services. Service provider will credit amount to the content provider and the peer. The billing is also raised from the user by using piggybacking when the next advertisement request is triggered from the user mobile device.
VI. CONCLUSION
We developed a privacy preserving mobile advertising system, where we considered a UN trusted ad-server and users who do not trust each other with their interest information. The architecture relies on cooperative behavior among nodes to request ads and distribute them to each other, and to implement a mixing algorithm to hide the interests of users from each other and their identities from the server.