Introduction
As information technology is accessible to more people, government and commercial organizations feel the challenge of scaling their services for widening user communities and enlarging their market shares while reducing costs for doing so, increased demand is there of software applications for providing few features like richer application with end-to-end functionalities, combinational usage of existing systems and software applications in adaptive ways and reduction of involvement of human to human end-users. Distributed multi-agent system (MAS) web services and technologies satisfy users by their modularity and recombining them for forming new applications. Multi-agent system is developed to implement particular test applications. MAS is formed on assumption that this would be operating in open world.
Characteristics of Environment of Multi-agent System
Partially Accessible
Environment of Multi-Agent System could be considered as accessible if agent could obtain accurate, complete and latest information about the state of environment at every point. The environment of MAS is accessible fully to system agent. State of a system is known every time. However, at some point of time, if any waste appears in building and no e-Searchers are available there near it, at that point, state of that environment is inaccessible to other non-system agents. Hence, the environment of MAS is partially accessible.
Static
As per the system of multi-agent, the environment of MAS is static. There are no changes in it even if the agents within the system are acting.
Discrete
Distinction of environment of multi-agent system with respect to continuous or discrete depend on three essential factors: the state of the environment of MAS, the process in which time is maintained and the actions and percepts of the system’s agent. Environment of MAS is discrete as there exists limited number of actions and perceptions, limited number of city’s possible states and time is maintained in discrete way.
Deterministic
These exists three main factors which affect the environment of MAS such as discovering of waste, collection of waste and waste disposal. Certainty is there for resulting state after any action is performed. Every action within the system has assured effect on the system’s environment, hence it can be stated as deterministic.
Non-episodic
Action in each episode in non-episodic environments could affect every future decisions. Current episode might depend on actions taken from previous episodes. Hence, this environment is non-episodic.
Agent Architecture and Properties
Logic-Based Subsumption Architecture
Architecture
Logic-Based Subsumption Architecture (LSA) is composed of layers which are correspond to effectors and sensors. These layers work asynchronously and concurrently with each other. These layers are provided with theorem prover and running concurrently their respective processing loop. Thus, deterministic environment is good for this architecture.
Properties
- Pro-activeness: These agents are capable of exhibiting objective-directed behaviour through taking initiative for satisfying the design objectives.
- Reactivity: These agents are capable in perceiving their environment and responding in timely fashion for changing which takes place in it for satisfying the design objectives.
- Social Ability: These agents are capable to interact with other available agents for satisfying the design objectives.
Reactive Architecture
Architecture
Reactive architecture is used for building robust and responsive distributed systems depending on asynchronous passing of message. This architecture focuses on quick response for change detection in environment, preserve changes in environment to response in timely manner that consist of a set of effectors and sensors. Dynamic environment is suitable for this architecture.
Properties
- Responsiveness: Agents should respond in timely manner consistently.
- Resilient: Agents should be responsive in case of any failure.
- Elastic: Agents needs to be responsive under different workload.
- Message Driven: Agents possess asynchronous messaging. While requesting of data, agents do not wait for response, callback is registered. When there is availability of data, the data is pushed into callback method.
Deliberative Architecture
Architecture
Deliberative Architecture is type of architecture which is mainly used in simulations of multi-agent system. This architecture focuses for a long-term planning of action centred on set of objectives that contains explicit symbolic links of world, makes decision by symbolic links. Deterministic environment is best for this kind of architecture.
Properties
- Desired States: Agents’ few states needs to be counterfactual in sense to refer to hypothetical future goals or actions.
- Hypothetical situation: Deliberative agents need representation by compositional semantics.
- Planning: Agent needs to plan for achieving the objective.
- Scheduling: Deliberative agents need to schedule their actions for achieving their objective.
Hybrid Architecture
Architecture
Hybrid Architecture is the architecture which exhibits both discrete and continuous behaviour, that is, an architecture which could both jump and flow. Discrete environment seems to be best for this architecture.
Properties
- Atomic: Each behaviour by agents in this architecture, associated MLS learn from environment through automate process.
- Social: Hybrid mechanism of behaviour co-evolution is applied to behaviours of every agents for sharing and operating acquired knowledge among them.
- Individual: Agent self-configures the internal structure, arbitration and hierarchy od behaviours by evolutionary process.
Belief-Desire-Intension (BDI) Architecture
Architecture
Belief-Desire-Intension (BDA) Architecture possess their roots into philosophical tradition to understand practical reasoning which is process to decide, each moment, which actions are need to be performed in furtherance of objectives. Thus, continuous environment seems best for this architecture.
Properties
- Intention Persistency: The agents track success of all their intentions, and try again if there is failure in their attempt.
- Intention Satisfiability: The agents believe that there is possibility in their intensions. They believe some way is there for bringing about their intention.
- Inconsistency in Intension-Belief: These agents have no believe that their intensions would be brought about. Agents’ irrational would be there for adopting an intension.
- Incompleteness in Intention-belief: There is no believe in agents that their intention could be achieved. It might be understood as the agents’ rational behaviour.
- Intension side-effects: The agents do not intend every side effects that are expected for their intentions.
Cognitive Architecture
Architecture
Cognitive Architecture refers to both theory of computational instantiation and of structure of human mind. Main objective of cognitive architecture is summarizing several cognitive psychology results in inclusive computer model. This architecture suits static environment.
Properties
- High-Level Cognition: Agents possess abilities like deep comprehension, problem solving and abstract reasoning. This could be contrasted with machine learning.
- System-Level Approach: Integrated, comprehensive cognitive architectures give seamless integration among several cognitive functions. All these functions are mutually supportive.
- Structured Representations: Every skills and knowledge needs to be encoded in integrated, uniform manner which reflects data’s logical structure.
- Links with Human Cognition: While copying how brains get intelligence are not needed by AI designs, crucial human cognition’s characteristics needs to be implemented for practical, successful intelligence engine.
Conclusion
This paper identifies the environmental characteristics of multi-agent system. The different agent architectures of MAS are discussed in this paper. In this paper, the multi-agent system is analysed. As seen in this paper, MAS are distinctive software architecture that provides solutions to computational problems’ specific family. It has features which make it unique discipline and promising to guide that large-scale multi-agent systems’ reliability could be offered by it. This paper also discusses about the agent properties of multi-agent system. MAS should be made more accessible for system designers and provide with means to consider MAS as prospective solutions for computational problems.
References
1 Tapia, D. I., Fraile, J. A., Rodríguez, S., Alonso, R. S., & Corchado, J. M. (2013). Integrating hardware agents into an enhanced multi-agent architecture for Ambient Intelligence systems. Information Sciences, 222, 47-65.
2 Kumari, S., Singh, A., Mishra, N., & Garza-Reyes, J. A. (2015). A multi-agent architecture for outsourcing SMEs manufacturing supply chain. Robotics and Computer-Integrated Manufacturing, 36, 36-44.
3 Antzoulatos, N., Castro, E., Scrimieri, D., & Ratchev, S. (2014). A multi-agent architecture for plug and produce on an industrial assembly platform. Production Engineering, 8(6), 773-781.
4 Twardowski, B., & Ryzko, D. (2014, August). Multi-agent architecture for real-time big data processing. In 2014 IEEE/WIC/ACM International Joint Conferences on Web Intelligence (WI) and Intelligent Agent Technologies (IAT) (Vol. 3, pp. 333-337). IEEE.
5 Xi, J., He, M., Liu, H., & Zheng, J. (2016). Admissible output consensualization control for singular multi-agent systems with time delays. Journal of the Franklin Institute, 353(16), 4074-4090.
6 Banerjee, S., & Hecker, J. P. (2017). A multi-agent system approach to load-balancing and resource allocation for distributed computing. In First Complex Systems Digital Campus World E-Conference 2015 (pp. 41-54). Springer, Cham
7 Lewis, F. L., Zhang, H., Hengster-Movric, K., & Das, A. (2013). Cooperative control of multi-agent systems: optimal and adaptive design approaches. Springer Science & Business Media.
8 Yu, Y., El Kamel, A., Gong, G., & Li, F. (2014). Multi-agent based modeling and simulation of microscopic traffic in virtual reality system. Simulation Modelling Practice and Theory, 45, 62-79.
9 Kouvaros, P., & Lomuscio, A. (2013, May). Automatic verification of parameterised multi-agent systems. In Proceedings of the 2013 international conference on Autonomous agents and multi-agent systems (pp. 861-868). International Foundation for Autonomous Agents and Multiagent Systems.
10 Araujo, F., Valente, J., Al-Zinati, M., Kuiper, D., & Zalila-Wenkstern, R. (2013, May). Divas 4.0: A framework for the development of situated multi-agent based simulation systems. In Proceedings of the 2013 international conference on Autonomous agents and multi-agent systems (pp. 1351-1352). International Foundation for Autonomous Agents and Multiagent Systems.
11 Yu, Y., El Kamel, A., & Gong, G. (2013, April). Multi-agent based architecture for virtual reality intelligent simulation system of vehicles. In 2013 10th IEEE INTERNATIONAL CONFERENCE ON NETWORKING, SENSING AND CONTROL (ICNSC) (pp. 597-602). IEEE.
12 Dou, C. X., & Liu, B. (2013). Multi-agent based hierarchical hybrid control for smart microgrid. IEEE transactions on smart grid, 4(2), 771-778.
