DATA RETRIEVAL IN ADHOC NETWORKS By Aviraj Sinha CSE5343

Ad-Hoc: Definition Revisited Definition: network based on personal connections for a specific task Each computer becomes a client, server, and router. Ad-hoc is a type of P2P network Can store data that can be mined (decentralized retrieval)

P2P

History of P2P Napster Spotify DropBox BitTorrent This Photo by Unknown Author is licensed under CC BY-SA This Photo by Unknown Author is licensed under CC BY This Photo by Unknown Author is licensed under CC BY-SA

Reason for Effectiveness Pieces of file can be simultaneously received The client that receives it can be used as a nodes Extremely cheap, redundant scalability Self-sufficient and efficient

Importance of Ad-hoc networks Users only private network allows efficiency of hops Capable of going off-grid (Legal implications) Reliably redundant for police and military

Potential VANET Cars will be self driven and will communicate with each other to prevent traffic jams and crashes TETRA radio for police—much better solution to thrashing in central radio systems

VANET

More Potential MANET- continuously self-configuring, physical infrastructure-less network Ease of networking on PAN, Far future use on multiple hierarchies of LAN and MAN

Set Backs Rapidly changing topology Network partitions and security Higher error rates, collision interference, bandwidth constraints, Power limitations

Data Retrieval Algorithm Characteristics WANET routing algorithms are different from traditional routing algorithms. Data retrieval requires huge amounts of data from decentralized storage. Asynchronous not keeping the time because it is too big, Fault tolerant-- the parts of data change while the algorithm is running

Searching Categories: Blind and Structured Search is blind –each node does not have information to where data needed is located. Depth first search (DFS) Or structured where the search is routed. Exact local algorithms

More Complex Methods– Intelligent Searching Mechanisms Neighbor Profile Relevancy Rankings pSearch, a non-flooding system using Latent Semantic Indexing Data overlays– using in depth local knowledge Genetic algorithms (Machine Learning)

Indexing Strategies Posting lists for carefully chosen indexing term combinations Statistically bounded number of their top-ranked elements. Local vs Global index Gossiping--global inverted index which is partially constructed by each node propagated to the rest of the network

All information shared, simple routing Purely Central Can use Normal Routing Protocols Requires More complex Algorithms

Hybrid technique Local indices contain the ”direction” to global index Compound Routing Indexes (CRI), Hop-Count Routing Index (HRI) and Exponentially aggregated RI (ERI) Different topologies (tree, tree with cycles) Uses Bellman Ford or Distance Vector Routing Algorithm

Final Take away What P2p is Strengths and Weaknesses Variety of Architectures and Algorithm Classification