[attachment=12553]
1. INTRODUCTION
A new and growing technique in the world of the Internet is the Peer-to-Peer architecture, usually referred to as P2P, in which each node on a network is assumed a peer for other connected nodes to have node-to-node data communication and resource sharing. As the number of Internet users is increasing and their knowledge of Internet resources is developing, a more powerful and convenient tool to communicate is in demand. The P2P technique may provide such a tool.
1.1 The Problem
Most of the traditional LAN (and later the Internet) architectures have been based on the client/server concept, in which one or more computers (servers) on the net are designated to serve the other computers (clients). Although this architecture is still being used widely, it cannot fully satisfy the requirements of communication between two objects on the Internet. Since a server must always be present in the client/server architecture and a client can only contact with the server, direct communication between two clients is difficult. However, direct communication is important for many applications. For example, a group of workers working on the same project may need a more “decentralized” communication environment, in which they may better share each other’s resources, help (serve) each other, and so on, without doing everything through a central server. P2P architecture is different from client/server architecture. In P2P, users can communicate with, request help from, and serve each other without a central server. For many applications, P2P is more suitable and more efficient than the client/server architecture. Hence, P2P architecture has started to be widely adopted by designers and developers. The number of programs that are based on P2P architecture is increasing because quite often users prefer to talk to each other more freely and to be less dependent on server for resources and services.
The P2P technology is still in its early developing stage. The current P2P systems cannot satisfy the growing needs for an advanced and powerful communication environment between any two nodes on the net. In the existing P2P systems users have to be responsible for finding resources by themselves and to decide the way in which a particular task can be done. That means that the quality and efficiency of communication and data sharing largely depends on users’ knowledge and their individual experience of P2P applications.
1.2 Paper Objectives
The goal of this paper is to overcome the limitations and deficiencies that exist in the current P2P systems. To do so, we developed a heuristic approach to achieve the goal. We apply Autonomous Agents as the main technology in developing an Advanced P2P Architecture, in which most of the characteristics of an ideal P2P have been considered. The Advanced P2P Architecture has the strengths of both virtual P2P and pure P2P.
The following are the major objectives of the Advanced P2P Architecture :
 Designing a Peer to Peer architecture on the Internet that covers most of the characteristics of an ideal P2P including:
 Community of peers who have pure and direct data communication
 Virtual server who united the peers to find each other
 Dynamic switching between the two modes of Virtual and Pure P2P when it is needed to take advantages of both architectures in one system.
 Comprehensive Peer-to-Peer communication on the Internet Resources and services sharing among users
 Giving a server-independent identity to the users on the Internet
 Internet resource mapping
 Implanting efficient and Autonomous Agents into Virtual P2P to improve P2P architecture with the following advanced features:
 Auto-interacted agents to work and consult with each other in problem-solving
 Agents who learn from the results of previous requests and conclude new results
 Guided search
 Agents who can do their job even when a user is not present
 Locating resources and service by agents autonomously
1.3 Broad Overview of Paper
In Chapter 2, the important concepts of distributed systems and Internet communication, and the current technologies and techniques are discussed. Advantages and disadvantages of these techniques are mentioned as well.
Later in Chapter 3, a heuristic design is presented. The desired features of the Advanced P2P Architecture are described and features of Autonomous Agents are discussed.
Finally, in Chapter 4, the paper is concluded and the major contributions of this research are summarized.
2. BASIC BUILDING BLOCKS
2.1 Distributed systems and network architectures
There are many different types of architectures for creating a distributed system but all can be defined in this classification:
Client/Server
 Single Sever
 Multi-Servers
Peer to Peer
 Pure P2P
 Virtual P2P (Combinational)
 Advanced P2P
2.2 The Traditional Architecture of Client/Server
Most of the current applications on LANs and the Internet are based on the client/server architecture or Web server/Browser architecture. Advantages and disadvantages of this traditional architecture, in terms of resource sharing and communication, should be considered when designing a revised and advanced P2P model.
The following are some of the salient features of the client/server architecture that an ideal P2P system should have:
 A server is able to provide an updated list of connected users for each client.
 Users who log into a particular server form a united community.
 A server can monitor and filter (if needed) the communication of its users.
 A server can keep the transferring data temporarily for the offline users (such as emails).
 A complete event recording can be performed through the entire P2P community by the Server.
 Public task processing is possible (i.e. public messages or inquiries).
 A server may help its users know other users who might haven’t met each other before.
Systems, based on server/client or master/slave, suffer from some problems. The most obvious are:
• Resources are limited to server resources.
• Users (clients) are not aware of resources (and services) that are available on other clients’ computers.
• Users cannot directly access the resources (and services) on other users’ computers.
• A server usually has to keep a huge amount of data.
• A busy server has to process a considerable number of jobs simultaneously and this may slow down services.
• Users always have to talk only with the server and it is boring!
2.3 Peer to Peer (P2P) Architecture
P2P architecture has been introduced as a solution for many of the current deficiencies in client/server-based applications. What is P2P architecture really? P2P is a new concept in Internet communications in which a node can act as a server for other nodes and also be a client to use other nodes’ resources and services. In a P2P architecture a computer connected to a network (or the Internet) can be a peer to another on the net. It uses and shares the latter’s resources and services and vice versa. P2P architecture itself can be designed and implemented in two types, Pure P2P and Virtual P2P. In pure P2P, two peers communicate with each other directly without interference of a third party. Pure P2P has the advantages of higher performance. In Virtual P2P, peers communicate with each other through a transparent virtual server. It is easier for virtual P2P users to obtain information about other P2P users and the resources/services within the user community.
In Figure 2-1, client/server and the two different types of P2P architectures are showed and compared.