1. 1 COACS: A Cooperative and Adaptive Caching System for MANETs Hassan Artail, Member, IEEE, Haidar Safa, Member, IEEE, Khaleel Mershad,Zahy Abou-Atme, Student Member, IEEE, and Nabeel Sulieman, Student Member, IEEE IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. 7, NO. 8, AUGUST 2008 Present: I-Wei Ting Date: Sep. 17, 2008 Manuscript received 8 June 2006; revised 12 Mar. 2007; accepted 23 Jan. 2008; published online 28 Jan. 2008.

2. 2 Outline Related work  Data discovery under ICP (Internet Cache Protocol) Proposed protocol  Election of Query Nodes (QN, Proxy) in MANETs Life time, Battery, Bandwidth, Memory  Data discovery phase  Management of the QN and Caching Node Performance Evaluation Analysis

3. 3 Internet Cache Protocol The ICP protocol was designed to be lightweight in order to minimize round-trip time between caches. It is intended for unreliable but quick connections, using short time-outs before a cache starts to retrieve an object on its own. UDP is commonly used as delivery protocol.time-outsUDPdelivery protocol The ICP protocol is described in RFC 2186, its application to hierarchical web caching in RFC 2187.RFC 2186 hierarchicalRFC 2187 Web proxies that support ICP include:  Squid cache  Microsoft Proxy  Cisco Content Engine  ProxySG

4. 4 Internet Cache Protocol (1/6) A BCD FEG S sibling Server Client How to find a data object (web page)? Case 1

5. 5 Internet Cache Protocol (2/6) A BCD FEG S sibling Server Client Case 2 1 2

6. 6 Internet Cache Protocol (3/6) A BCD FEG S sibling Server Client Case 3 1 22 3 4

7. 7 Internet Cache Protocol (4/6) A BCD FEG S sibling Server Client Case 4 1 22 34 5

8. 8 Internet Cache Protocol (5/6) A BCD FEG S sibling Server Client Case 5 1 22 3 4 4 5 6 7

9. 9 Internet Cache Protocol (6/6) A BCD FEG S sibling Server Client Case 6 1 22 3 44 56 7 8

10. 10 Property Comparison InternetMANETs EntitySpecific nodes (Proxy server) Any nodes LocationFixedMobility Leave/JoinNot frequentFrequent Protocol OverheadLessMuch

11. 11 Election of Query Nodes (1/6) A BCD FEG S RN :Requesting Node Initial: No QD (Query Directory; Proxy server) Each RN only cache its requested data object. Then, query index is sent to the nearest QD

12. 12 Election of Query Nodes (2/6) A BCD FEG S RN :Requesting Node Find out the first QD (Query Directory; Proxy server) COACS Score Packet (CSP) Traverse all nodes in the networks sequentially (need to adopt routing table, DSDV)

13. 13 Election of Query Nodes (3/6) A BCD FEG S RN :Requesting Node Send QD Assignment Packet To the E (highest score)

14. 14 Election of Query Nodes (4/6) A BCD FEG S RN :Requesting Node Find out Other QDs based on the # of QDs and score QD1

15. 15 Election of Query Nodes (5/6) A BCD FEG S RN :Requesting Node QD1 QD1 find other QDs (ex:3) high. score ACK:OK

16. 16 Election of Query Nodes (6/6) QD1E QD2S QD3D A BCD FEG S QD1 QD Information packet Broadcast QD list to all nodes

17. 17 Data discovery phase QD1E QD2S QD3D A BCD FEG RN :Requesting Node K S

18. 18 Case 1: QD E has an entry for the query A BCD FEG RN :Requesting Node K S

19. 19 Case 2: QD forwards the request to the nearest QD QD1E QD2S QD3D QD1E QD2S QD3D A BCD FEG S RN :Requesting Node QD1E QD2S QD3D K

20. 20 Management of QD and CN A new QD is added to the system when a query needs to be cached but no QD agreed to cache it.  The last QD to receive the caching request will initiate a CSP When a QD receives a query and related CN is offline  Delete associated entries and forward the query to the original server CN offline  QD detect and remove related entries Lookup routing table (proactive routing) QD ack CN periodically (on-demand routing) QD offline  First node detect Run “add a new QD”

21. 21 Packet types

22. 22 Comments Node join  The available caching space can not be utilized efficiently

23. 23 Simulation parameters 1000m*1000m Nodes: 100 Cache size: 200Kb Trans. Range: 100m DSDV routing protocol Random Waypoint mobility model Speed:0.01~2, 10~20 m/s Query interval: 10 seconds DB: 10,000 items, 10Kb/item Number of QD nodes: 7

24. 24

25. 25

26. 26 Analysis 1. Expected Number of Hops between Two Nodes  E[H] = 0.521 * (a/r) expected minimum number of hops between any two nodes in the network, Ex: rectangular topology a=1000m 2, r=250m, 0.521*4=2.084 2. Expected Number of Hops within the System of Query Directories

27. 27 Analysis 3. Expected Number of Hops to the External Network 4. Query Directory Access and Delay T in : the delay for transmitting packets between nodes inside the network T out : the delay for accessing a node outside the network (data source)

28. 28 Analysis 5. Determining the Maximum Number of Query Directories

29. 29 Analysis 6. Load Balancing on Query Directories