Research and Reviews - International Journals

International Journal of Innovative Research in Computer and Communication Engineering

Menu

ISSN ONLINE(2320-9801) PRINT (2320-9798)

All submissions of the EM system will be redirected to Online Manuscript Submission System. Authors are requested to submit articles directly to Online Manuscript Submission System of respective journal.

A Survey on Optimizing Search Topology in Overlay Networks

C.Priyanka1, C.A.Yogaraja2, Dr.K.Deeba3
  1. PG Scholar, Department of Computer Science and Engineering, Kalaignar Karunanidhi Institute of Technology Coimbatore, India
  2. PG Scholar, Department of Computer Science and Engineering, Kalaignar Karunanidhi Institute of Technology Coimbatore, India
  3. Associate Professor, Department of Computer Science and Engineering, Kalaignar Karunanidhi InstituteofTechnology Coimbatore, India
Related article at Pubmed, Scholar Google

Visit for more related articles at International Journal of Innovative Research in Computer and Communication Engineering

International Journal of Innovative Research in Computer and Communication Engineering

Abstract

Overlay network both structured and unstructured have a massive raise in today’s market. These networks can be constructed very efficiently without precise rules and therefore considered suitable to the Internet environment. But, the search techniques used in these networks usually perform poorly with a large network size and oblivious to the physical network topology that introduces wide-area network traffic. To optimize the search performance in overlay network different strategies of search patterns are used. These searches are done with various topologies in order to expedite the search process through self-organizing the P2P network into a small world. Both theoretical and experimental results are been analysed also effectiveness and efficiency of various approaches in comparison are discussed.

Keywords
Overlay Networks, Optimizing, Peer topology, Search

INTRODUCTION
Overlay Networks provides various services as file sharing, instantaneous messaging and other popular network applications that rely on P2P technology. Peer-to-peer (P2P) networking eliminates the need for central servers, allowing all computers to exchange and share resources as equals. This growth of such applications introduces internet traffic up to 20 per cent as in prior studies [1] can have significant impact on the underlying network. Searching an object is an essential part in several applications where in the popular protocols such as Gnutella are unstructured and peers participating interconnect randomly through message flooding by enquiring a peer and broadcasts the message to its neighbours that have end-to-end connections(say p1 sends a query). Those broadcasted message are connected with a positive integer time to live (TTL) value. On receiving a message by the peer (say p2) decreases its TTL value associated with it by 1 and then relay the message with the updated TTL value associated to its neighbours. Upon receiving a query a peer p2 try searching in local store to see if it is possible to provide the objects that are requested by peer p1. It can respond either by sending the objects that are been requested directly or can return objects to the overlay path where the query message traverses from p1 to p2. This phenomenon of searching.
In Semantic Overlay Networks (SONs), peers that are socially similar are organised into groups. SONs, while being highly flexible, improve query performance and promise high degree of peer autonomy [3]. Additionally, this method offers better support of semantics due to their ability to provide mechanisms for estimated range, or text queries, and highlight peer independence.an object takes place in most of the overlay networks.
Rather than this type of search topology another approach given to the search enhancement was RefNet method, which focused on the effect of network clustering that, finds ability of relevant information sources of the decentralized search [4].
In highly dynamic P2P systems, when a selected peer leaves, the whole path fails. Unfortunately, such a collapse is often difficult to be known by the initiator. Therefore, a randomly-assigned path is very variable, and users have to frequently inquiry the path and retransmit messages, to address this issue a non-path-based anonymous light weight P2P protocol called Rumor Riding (RR) was anticipated. Where RR provides a high degree of anonymity and outperforms existing approaches in terms of reducing the traffic overhead and processing latency. The overall idea of this paper is that the statistical patterns over locally shared resources of a peer can be explored to guide the distributed resource discovery process and therefore enhance the overall resource discovery performance in unstructured peer to peer networks.

RELATED WORK
In this section we primarily discuss about P2P networks that aims to exploit the similarity of participating peers. pSearch[6] and SSW[7] are content based P2P networks providing semantic search. Similar to most overlay networks based on distributed hash tables (e.g., Chord[8]) in pSearch and SSW, the peers those participating need to maintain foreign indices, that is, the indices of the object stored in the distant peers. To locate an object, a requesting peer routes a message toward peer responsible for the key subspace where the object is indexed. In contrast, we can construct unstructured peer to peer networks where the participating peers need not to organize themselves into a secure, deterministic topology group, considerably reducing the maintenance overhead of overlay topology. Unlike pSearch [6] and SSW[7], the peers in the network hosts the objects of interest and maintain no foreign index, eliminating storage bandwidth overheads for publishing and managing such index. However, due to space complexity, we focus on optimizing searching protocol. While maximizing the coverage of a query message may not guarantee the efficiency and efficiency of searches.
In an unstructured peer to peer network environment, the concept of anonymity, many approaches have been proposed to support anonymous communications. The prior methods in the second generation such as Onion Routing, Tor, APFS, Shortcut Protocol provides P2P mutual anonymity with a reduced response delay [2]. When receiving a packet, a peer has two options as such forwarding the packet to a randomly chosen peer or directly sending it to the destination peer. All these above methods have different mechanisms focusing on the peer search improvement and managing the overlay topology by different methods.

EXPLOITING PEER SIMILARITY
Consider any given unstructured P2P network G = (V, E), where V is the set of peers that are participating, and E is the set of overlay connections linking the peers in V. The peers in G may be interrelated randomly. Where G should satisfy the following strategies to exploit the peer similarity such as:
A. High Clustering:
Any peer in V should connect max peer in V and the neighbours are selected in top most cadets of V.
B. Low Diameter:
Consider two different peers in V, such that both peers should have at least one overlay path existing between them, with the hop count almost small as possible.
C. Progressive:
There should be a path existing between the peer who issues the query and the peer which resolves the query.
The above A, B, C should be maintained effectively in the following overlay search topologies. Exploiting the peer similarity is taken by;
Definition:
Let V be the set of peers participating in an Overlay network. The peer semblance function measures the degree of the similarity between any two peers, u?V and v?V in the system where u and v is any two peers that participate.
Consider Figure 1 as Graph G = (V, E), where V={1,2,3}. Peers 1,2 and 3, respectively, host set of objects O1={a}, O2={a,c}, O3={a,b,c}. Any two peers u and v have an edge in E if both peers share at least one common object. That is, F(u,v) =?Ou ∩ Ov ? Ou U Ov?,the value nearby an edge (u,v) indicates F(u,v). F(u ,v) is defined as the inverse of the cosine angle of two summarized latent semantic vectors representing any two peers u and v in a P2P network, where every element in a summarized vector for any peer say i calculates the total frequency of the corresponding keyword appearing in the data items stored in i. The following table debits the notations that are frequently used in exploiting the peer similarity.
This type above mentioned similarity function is used for search [1].

P2P NETWORK COMPARITIVE STUDY
In this section various methods that are used previously for search protocol with various topologies are discussed. Chord, Napster, Guntella are some of the famous search protocols in market. In Chord [8], the problem in peer to peer systems is opposed by efficiently locating the node that stores a particular data item. Chord adapts efficiently as nodes join and leave the system, and can respond to queries even if the system is unceasingly changing. The performance investigation shows that Chord scales well with the number of nodes, recovers from large numbers of instantaneous node failures and joins, and answers most lookups correctly even during recovery. The other popular search protocols such as Napster and Gnutella are compared on the basis of their characteristics such that, there is significant heterogeneity and lack of cooperation across peers participating in these systems [9].
Flooding is an essential building block of unstructured peer-to-peer (P2P) systems. To improve the performance from flooding a novel method of Clustella, a novel semi-structured P2P architecture with bounded peer degree[10]. It decomposes the network into different clusters, allowing peers to quickly find those neighbours which contribute much to their routing efficiency both in static and dynamic environments. Later another method to overcome the flooding was Rumor Riding (RR), a lightweight and non-path-based mutual anonymity protocol for decentralized P2P systems. Peers participating in unstructured networks interconnect randomly, rely on flooding query messages to discover objects of interest introduces Network traffic uses a sower to distribute and redistribute the query. The performance analysis shows that RR provides a high degree of anonymity and outperforms existing approaches in terms of reducing the traffic overhead and processing latency [2].
In peer to peer next section upon content based retrieval methods like pSearch, SSW where the next problem raised. pSearch distributes document indices through the P2P network based on document semantics generated by Latent Semantic Indexing (LSI) [11].

ENHANCED SEARCH PROTOCOL
In order to provide an enhanced support to this type of peer environment for both structured and unstructured environment the balancing of peer search topology can be studied. The major study is on research of structured peer to peer systems where the nodes are often being homogenous and sometimes to be heterogeneous. While typically there are several load balancing schemes have been proposed are typically ad hoc, heuristic based, and localized. For a structured network HiGLOB, for global load balancing in structured P2P systems was proposed. Each node contains a histogram manager which maintains a histogram that reflects a global view of the distribution of the load in the system also load-balancing manager redistributes the load whenever the node becomes overloaded or under loaded. This outperforms Skip Graph, BATON, and Chord in balancing the peers [15]. Pastry[12] another method where it can efficiently support a variety of peer-to-peer applications. Being decentralized it is widely used for large-scale peer-topeer systems which out performs Napster, Gnutella and also FreeNet.
Most P2P networks based on distributed hash tables where, different problem on balancing load in p2p systems can be solved by providing a DHT abstraction distribute objects among “peer nodes” by choosing random identifiers for the objects. Designing load-balancing algorithms that uses the notion of “virtual servers” is able to balance the load within 80% of the perfect value, while the most complex scheme is able to balance the load within 95% of the optimal value [13]. In an active environment being the nodes to be heterogeneous, of the nodes varying rapidly an algorithm was presented. The simulation results show that in face of rapid arrivals and departures of objects of widely changing the load, load balancing for system utilizations as high as 90% while moving only about 8% of the load that arrives into the system [14]. Similarly many such load balancing techniques can be studied such imposes the enhancement of peer to peer network. Those of which simulation results can be compared with the existing search topologies where comparative analysis of different search method.
Here certain method of which the load balancing can be done are discussed. The P2P networks can also be based on maintaining the balance information of peers in Routing table. The differences in both tables are such DHT offers security than Routing table. In concern of search enhancement alone the efficiency is better rated in routing table than DHT. A comparative analysis can be done on both efficiency rating between both maintaining balance information on DHT and routing table.

CONCLUSION AND FUTURE WORK
In this paper the study of Overlay network is been given with different methods that have been proposed in prior with their performance comparison on basis of how efficiently the query for search protocol in both the structured and unstructured environment are discussed. Peer environments being decentralized are been used in rapid growth of development of various applications. The comparative analysis of different strategies in peer environment would be rather interesting in designing various topologies.
Our Future work is on the contribution of a topology that is entirely applicable for both structured and unstructured environments. It is also interesting in maintaining heterogeneity throughout supporting dynamic peers. Further the wide area network traffic, storage space, and computational capability that occur in various clustering based search technique is also been studied.

Tables at a glance
Table icon
Table 1
 

Figures at a glance
Figure 1
Figure 1
 

References