THE COMPRESSION OF PIT WITH BLOOM FILTER IN CCN Zhaogeng Li, Jun Bi, Sen Wang, and Xiaoke Jiang Asia FI Workshop in Kyoto, 2012 Sho Harada Park Lab Nov 29th, 2012 OUTLINE 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. Introduction CCN (Content Centric Networking) Bloom Filter Architecture Problem United Bloom Filter Error Handling Experiments Conclusion Reference 2 1. INTRODUCTION CCN was developed to solve many network problems that is being occurred from increasing traffic. It is one of the most promising architectures as a Future Internet architecture. CCN router uses three tables that store data. This proposal enables us to compress the size of the table. 3 2. CCN (CONTENT CENTRIC NETWORKING) Packet Interest Packet : Used to request a content. Data Packet : Used to send the content. CCN router CS (Content Store) : Cache contents. PIT (Pending Interest Table) : Record name and face to define where to forward Data Packet. FIB (Forwarding Information Base) : Record face to decide where to forward Interest Packet. 4 2. CCN (CONT.) 5 3. BLOOM FILTER 6 3. BLOOM FILTER (CONT.) 7 3. BLOOM FILTER (CONT.) 8 3. BLOOM FILTER (CONT.) 9 3. BLOOM FILTER (CONT.) 10 4. ARCHITECTURE Bloom Filter is introduced in PIT. Content Name is converted by hash function and added to Bloom Filter of the appropriate face. 11 4. ARCHITECTURE (CONT.) 0 1 2 PIT Bloom Filter Face 00000000 0 00000000 1 00000000 FIB 2 Name Face Youtube/Video.mp4 1 12 4. ARCHITECTURE (CONT.) 0 1 Interest “Youtube/Video.mp4” 2 PIT Bloom Filter Face 00000000 0 00000000 1 00000000 FIB 2 Name Face Youtube/Video.mp4 1 13 4. ARCHITECTURE (CONT.) Interest “Youtube/Video.mp4” 0 1 2 PIT H( “Youtube/Video.mp4” ) = “01010101” Bloom Filter Face 01010101 0 00000000 1 00000000 FIB 2 Name Face Youtube/Video.mp4 1 14 4. ARCHITECTURE (CONT.) Data “Youtube/Video.mp4” 0 1 2 PIT H( “Youtube/Video.mp4” ) = “01010101” Bloom Filter Face 01010101 0 00000000 1 00000000 FIB 2 Name Face Youtube/Video.mp4 1 15 5. PROBLEM Data “Youtube/Video.mp4” 0 1 2 H( “Youtube/Video.mp4” ) = “01010101” H( “Youtube/Video2.mp4” ) = “00001111” PIT Bloom Filter Face 01011111 0 00000000 1 01010111 FIB 2 Name Face Youtube/Video.mp4 1 16 5. PROBLEM (CONT.) Data “Youtube/Video.mp4” 0 1 2 H( “Youtube/Video.mp4” ) = “01010101” H( “Youtube/Video2.mp4” ) = “00001111” PIT Bloom Filter Face 00001010 0 00000000 1 00000010 FIB 2 Name Face Youtube/Video.mp4 1 17 5. PROBLEM (CONT.) 0 1 Interest “Youtube/Video.mp4” 2 PIT H( “Youtube/Video.mp4” ) = “01010101” Bloom Filter Face 01110101 0 00000000 1 00000000 FIB 2 Name Face Youtube/Video.mp4 1 18 6. UNITED BLOOM FILTER Use two Bloom Filters in one face. Filter shifts active and inactive. When a Bloom Filter stops, it will be initialized. 19 6. UNITED BLOOM FILTER (CONT.) Time Filter 1 Filter 2 Filter 1 = “01010101” (Active) Filter 2 = “00000000” 20 6. UNITED BLOOM FILTER (CONT.) Time Filter 1 Filter 2 Filter 1 = “01010101” (Active) Filter 2 = “00000000” (Record) 21 6. UNITED BLOOM FILTER (CONT.) Time Filter 1 Filter 2 Filter 1 = “00000000” Filter 2 = “00111100” (Active) 22 7. ERROR HANDLING The result of experiment shows that the probability of false positive was less than 0.1 %. If an Interest Packet was dropped, the requester sends Interest Packet again. Data may be forwarded by false positive. But the Data Packet will be dropped by the next node. 23 8. EXPERIMENTS BF : 1MB 24 8. EXPERIMENTS (CONT.) Compression of PIT : 40% reduced Probability of False Positive : 0.027% 25 9. CONCLUSION Introducing Bloom Filter, the compression of PIT is realized. When we use Bloom Filter, we need to think of False Positive. ⇒ Experiment shows the probability of False Positive was only 0.027 %. Therefore, it will not make a big problem. We have only to deal with False Positive when it happens. 26 10. REFERENCE Zhaogeng Li, Jun Bi, Sen Wang, and Xiaoke Jiang, “The Compression of PIT with Bloom Filter in CCN”, Asia FI Workshop in Kyoto, 2012. 27 28