Multi-agent system (MAS) is a common way of exploiting the potential power of agent by combining many agents in one system. Each agent in a multivalent system has incomplete information and is in capable of solving entire problem on its own. Multi-agent system offers modularity. If a problem domain is particularly complex, large and contain uncertainty, then the one way to address, it to develop a number of functional specific and modular agent that are specialized at solving various problems individually. It also consists of heterogeneous agents implemented by different tool and techniques. MAS can be defining as loosely coupled network of problem solvers that interact to solve problems that are beyond the individual capabilities or knowledge of each problem solver. These problem solvers, often ailed agent are autonomous and can be heterogeneous in nature. MAS is followed by characteristics, Future application, What to be change, problem solving agent, tools and techniques used, various architecture, multi agent applications and finally future Direction and conclusion. Various Characteristics are limited viewpoint, effectively, decentralized; computation is asynchronous, use of genetic algorithms. It has some drawbacks which must be change to make MAS more effective. In the session of problem solving of MAS, the agent performance measure contains many factors to improve it like formulation of problems, task allocation, organizations. In planning of multivalent this paper cover self-interested multivalent interactions, modeling of other agents, managing communication, effective allocation of limited resources to multiple agents with managing resources. Using of tool, to make the agent more efficient in task that are often used. The architecture o MAS followed by three layers, explore, wander, avoid obstacles respectively. Further different and task decomposition can yield various architecture like BDI (Belief Desire Intension), RETSINA. Various applications of multi agent system exist today, to solve the real-life problems, new systems are being developed two distinct categories and also many others like process control, telecommunication, air traffic control, transportation systems, commercial management, electronic commerce, entertainment applications, medical applications. The future aspect of MAS to solve problems that are too large, to allow interconnection and interoperation of multiple existing legacy systems etc.
A Review on Intelligent Agent Systems
