Transcript ppt - MMLab
Ad Hoc Networking via Named Data Michael Meisel, Vasileios Pappas, and Lixia Zhang UCLA, IBM Research MobiArch’10, September 24, 2010 2011. 3. 13 Shinhaeng Oh ([email protected]) 1/22 CONTENTS • Background – Internet Protocol vs. Named Data • Existing Solutions for mobile networks – Ad-Hoc Networking over IP – Limitation of IP-Routing • New Direction for mobile networks – NDN for Ad-Hoc Networking – Design Example : LFBL • Conclusion 2/22 Introduction • TCP/IP and CCN Protocol Stacks – Replace packets with Data Objects or Interests – Replace Addresses with Names of Objects 3/22 Ad-Hoc Networking over IP 2 1 212.123.3.214 201.239.0.101 3 198.102.182.104 5 4 112.191.203.117 162.201.193.210 1. Each node is assigned an IP address 4/22 Ad-Hoc Networking over IP 2 1 212.123.3.214 201.239.0.101 3 198.102.182.104 5 4 112.191.203.117 162.201.193.210 2. Applications communicate by sending data to specific destination addresses 5/22 Ad-Hoc Networking over IP 2 1 212.123.3.214 201.239.0.101 3 198.102.182.104 5 4 112.191.203.117 162.201.193.210 3. When node move, determine a single best path to the given destination IP, and delivers data 6/22 Limitations of the IP-Routing (1) • Difficult to assign IP addresses (moving nodes) – IP addresses management is tightly controlled – It requires infrastructure support (e.g. DHCP) ad-hoc networks need infrastructure-free !! • In mobile, IP address is less meaningful – Wired networks, IP represent topology location – But, ad-hoc network do not have fixed location – Temporary unique identifier for device is needed SNU: 147.46.174.xx MIT: 18.9.22.xx 7/22 NDN for Ad-Hoc Networking (1) • Assign IP address to each nodes --No longer needs – To forward interests & data packets, – Nodes can use application data names directly forward or broadcast interest interest 8/22 Limitations of the IP-Routing (2) • Data is invisible in today’s IP-centric architecture source destination – It’s sub-optimal delivery • Accuracy of routing state maintained at each node Overhead to keep this state consistent --tradeoff – High node mobility – Constant movement in the aggregate at a large network 9/22 NDN for Ad-Hoc Networking (2) • Caching (traditional approach) – Ideally, each cached object has to be retrieved in its entirety from the same caching node. – But, images & audios & videos cannot fit within one packet – Transparent caching techniques work only in static network • Caching (NDN) – Intermediate node can forward to request node any part of file subsequent request 10/22 Limitations of the IP-Routing (3) • Receivers are in a better position to make forward decision than senders – In broadcast channel, nodes can hear the transmission – To keep all neighbors’ movement and connectivity changes will increase the routing table update overhead 11/22 NDN for Ad-Hoc Networking (3) • Interest packets can be forwarded multiple path – More than one direction returns the request data – A node can evaluate which path gives the best performance – Send future Interest for same data source in that direction – Remove critical dependency on pre-computed single paths 12/22 Design Example: LFBL • LFBL: Listen First, Broadcast Later • Uses a variation of NDN’s 3-way exchange – Name prefix announcements Response – Interest forwarding – Data return REQUEST Name of application data Response 13/22 Design Example: LFBL • LFBL: Listen First, Broadcast Later • Uses a variation of NDN’s 3-way exchange – Name prefix announcements – Interest forwarding – Data return ACK Destination 14/22 Design Example: LFBL • LFBL: Listen First, Broadcast Later • Uses a variation of NDN’s 3-way exchange – Name prefix announcements – Interest forwarding – Data return 2 Destination ACK 1 15/22 Design Example: LFBL • LFBL: Listen First, Broadcast Later • Uses a variation of NDN’s 3-way exchange – Name prefix announcements – Interest forwarding – Data return Destination ACK 1 2 16/22 Design Example: LFBL • LFBL: Listen First, Broadcast Later • Uses a variation of NDN’s 3-way exchange – Name prefix announcements – Interest forwarding – Data return ACK Destination 17/22 Design Example: LFBL • LFBL: Listen First, Broadcast Later • Uses a variation of NDN’s 3-way exchange – Name prefix announcements – Interest forwarding – Data return Destination ACK 1 2 3 18/22 Design Example: LFBL • LFBL: Listen First, Broadcast Later • Uses a variation of NDN’s 3-way exchange – Name prefix announcements – Interest forwarding – Data return ACK Destination 19/22 Performance Evaluation • Implemented LFBL in QualNet network simulator – Effect of % of mobile nodes – Move at a fixed rate of 30m/s (random waypoint mobility) various contents concurrently? 20/22 Conclusion • Frequent changes in topology had a direct impact on the performance of current protocols • Designed a new forwarding protocol: LBFL – For highly dynamic multi-hop wireless networks – Distributed forwarding capability with essentially no routing protocol • Through named data networking approach, – We can sketched out promising architectural direction to develop effective and efficient solution for ad-hoc networks 21/22 QnA 22/22 Related Work: DSDV, AODV • Destination-Sequenced Distance-Vector Routing (DSDV) is a table-driven routing scheme for ad hoc mobile networks based on Bellman-Ford algorithm – Each entry in the routing table contains a sequence number, they generally even if a link is present, odd used • For example the routing table of Node A in Network Destination Next Hop # of Hops Seq. number Install Time A A 0 A 46 001000 B B 1 B 36 001200 C B 2 C 28 001500