LENS LEveraging anti-social Networking against Spam (Introduction)

1. LENS LEveraging anti-social Networking against Spam (Introduction) MSc. Sufian Hameed Dr. Pan Hui Prof. Xiaoming Fu

2. Agenda • Introduction and Motivation • State of the Art • LENS • Experiments and Results

3. 1. Introduction and Motivation • Spam • Unsolicited bulk messages sent indiscriminately • Increased from 65% in 2005 to 81% in 2009 • 200 billion spams with avg size of 8Kbytes • Per day space consumption and bandwidth usage is 1,525,879 GB • Common Protection Techniques • Content-Based Filtering • Sender Authentication • Header-Based approach • Social Network Approach • Problems • False positives and negative • Spam already traversing the network

4. 2. State of the Art • Personal Email • a social network of friends in the cyberspace based on the emails exchanged between them • local clustering properties of social network classify emails • able to classify 53% of all the emails as spam or non-spam with 100% accuracy. • limited to offline analysis • 47% emails are left for other filtering techniques. • Reliable Email • Uses whitelist of friends and FoF to accept email • Accepts 85% of the emails and prevents 88% of false positives • Infrastructural overhead (public/private keys Attestation Server)

5. 3. LENS: LEveraging anti-social Networking against Spam • Anti-social networking paradigm, based on an underlying social infrasrtucture • Extend spam protection beyond social network • Prevent transmission of spam across the network • Receive all legitimate emails • Prevents all spam transmission • LENS consists of two parts • Formation of social network .i.e. community formation • Anti-social networking i.e. GK selection

6. 3.1 Community Formation

7. GK Selection Add to SKList Add to SKList Add to SignList SKList (SK, GKID, RNID) SignList (Signature[(CNID)Sign-SK, GKID, RNID ])

8. GK Selection – stage 1 CommLists 1 5 12 – F 6 – F 19 – F 17 – F 14 – F 11 – FoF – 12 10 – FoF – 6 20 – FoF – 19 18 – FoF – 17 16 – FoF – 17 13 – FoF – 14 15 – FoF – 14 3 – F 33 – F 36 – F 32 – F 31 – F 2 – FoF – 3 4 – FoF – 33 34 – FoF – 33 35 – FoF – 36 38 – FoF – 36 37 – FoF – 32 30 – FoF – 31 SKList 1 5 SK ,5, 1 SK ,5, 1 SignList 6 19 Sign[(19)SK, 5, 1] Sign[(6)SK, 5, 1]

9. GK Selection – stage 2

10. GK Selection – stage 3 Authentication Annonce

11. Email Processing

12. Email processing with LENS

13. 4. Experiments and Results Concerned in evaluating two things • Scalability • OSN Date (FaceBook and Flickr) • Effectiveness at accepting all the legitimate inbound emails. • Two real email traces (Enron and Uni-Kiel)

14. OSN Data • Interested in • # of GKs for receiving messages • Reachablity of recipient via GK

15. FaceBook • 4000 nodes • Community size 100-1500 • Numberof GKs • GKs between 56-880 • SKListentry in 76 bytes • 70 Kbytes in worsecase • Reachablityofrecipient via GK • Between 710K - 1.7 million (23-54%)

16. Flickr • 4000 nodes • Community size 100-1500 • Numberof GKs • GKs between 25-397 • SKListentry in 76 bytes • 28 Kbytes in worsecase • Reachablityofrecipient via GK • Between 682K-920K (39-54%)

17. Email Data Set • Enron • Contains data from mostly senior management of Enron. • Uni-Kiel • Data taken from log files of the email server at Kiel University over a period of 112 days.

18. Evaluations of Email Dataset • Email Acceptance • Number of GKs • Space Requirement • Message Overhead

19. Thank You