A mobile agent is a process that can transport its state from one environment to another, with its data intact, and be capable of performing appropriately in the new environment. Mobile agents decide when and where to move. Mobile Agent, namely, is a type of software agent, with the feature of autonomy, social ability, learning, and most importantly, mobility.
The middleware agent is a server process that is used to manage servers that are not running the product. The middleware agent can run on any node. You can install the middleware agent on a node that is running other middleware servers that were created outside of Web Sphere(R) Virtual Enterprise .Middleware agents have a framework with which they can communicate. The deployment manager can communicate with the middleware agents. The middleware agents must be at the same fix pack level as the deployment manager.
The remainder of this paper is organized as follows. Section II provides overview operating system middleware techniques. Section III describes the different models. Conclusions appear in Section IV.
os Techniques
Dynamic Quality of Service Middleware Agent
In past operating system is capable of handling the throughput requirements of the applications at the end of the network and it cannot guarantee the resources needed to application are provided. To overcome that problem developed a middleware agent called a dynamic Quality of service manager (DQM), that mediate application resource usage and to ensure that applications get the resources they need for its enough performance. Create an architecture in which there can be negotiation between applications, enforcement of applications, registered resource usage, and high resource utilization. Focused on supporting soft real-time processes on general-purpose operating systems.[1]
In order to give the policy decisions to the applications developed the execution level application execution model. In which each application is characterized by two numbers, the maximum CPU usage (Need to execute) and the maximum benefit (user specified indication). These execution level are application depended.
In this paper developed a prototype system consisting of a middleware Quality of Service manager called a Dynamic QoS Resource Manager (The DQM dynamically determines levels for the running applications based on the available resources and the specified benefit of the application.), a library of DQM interface and soft real-time support functions called the Soft Real-Time Resource Library (SRL allow the application define <Level, Resource usage, Benefit, Period>). DQM manager changes the level of each running application until all applications run without missing deadlines, system utilization and satiability.[1]
Parallel Multi Agent Middleware
Idea of Parallel Multi Agent System with a Session Agent that performs the dynamic service reconstruction according to the determination of user’s movements and service patterns as well as the reasoning of service procedures.
Introduce the idea of smart space, when multi users enter to smart space, a Session Management Function creates the user session and then registers and gather user profile and the service history. When multi users request the services and queries through heterogeneous networks and service domains, Session Agent first gathers the contexts based on authentication data and then reconstructs the differential services according to the service level and the device reconfiguration. To support the seamless service, we specify differential rules and then incorporate the parallel process into the Session Agent.[2]
Base for Building Secure middleware Mobile Agent
A layer between local operating systems and higher level agent platform middleware, it supports interoperability between agent platforms and between different implementations of AOS itself. Different requirements are design a middleware for mobile agent which are design a natural system, minimal code, aos reasonably efficient, not impose design limitation on the mobile agent middleware.
Agent operating system provide these functionalities; agent container which contain the Agent code, data, and meta-data, communication end point handle the Communication-related calls include creation and deletion of communication endpoints, provide the secure sharing of a single AOS instance and its resources, authentication model based on public key cryptography, provide the secure mobile migration through the secure Table of Contents of agent container. TOC data structure is available on MAM layer.[3]
Techniques
SOMA MODEL
SOMA is application of middleware agent. Soma is a Java-based platform that provides a layered service infrastructure for designing, implementing, and deploying Multi Agent based Internet applications. SOMA’s architecture consists of four layers mobility middleware, core services, Java virtual machine, and heterogeneous distributed system. The mobility middleware layer implements UVE, MVT, and VRM services.[4]
The core services layer includes communication, migration, naming, security, interoperability, persistency, and QoS adaptation services.[4]
AGILLA MODEL
Agilla model, apply on fire case study. This model addressing the destination of agent by using sensor, which get their location through GPS; rather than their ID. Agilla support the local tuple space rather than global tuple space. Agent can access this space remotely. Introduce the idea of rrdpg (remote probing group read) in which use the idea of multicast. Agilla maintain the list of neighbour agent list. In this model reduce the impact of message lose (divide the agent into tiny pack and migrate).[5]
M4WEBGIS
WebGIS middleware, called M4WebGIS, which encapsulates graphic user-machine interaction interface. WebGIS uses IBM Aglet as fundamental platform. Four modules of WebGIS is system interface, application interface, kernel part and extended function part. System interface are responsible for communication; application interfaces communicate with MapXtreme via map APIs; kernel part encapsulates several advanced features (advanced query, map control, other features, user define function); extended user defined parts interact with kernel via extended user interfaces.
There are six kinds of components in M4WebGIS: map data access component, map visualization component, map editing component, map spatial analysing component, practical function component and utility component.[6]
LIMONE MODEL
Limone is coordination model, facilitates rapid application development over ad hoc networks consist of mobile agent (logically mobile agents and physically mobile hosts). In Limone Agents are represented as ovals, agent owns a local tuple space (LTS), an acquaintance list (AQL), reaction registry, reaction list, and operation manager. Limone provides a beacon-based discovery protocol that informs each agent of the arrival and departure of other agents.[7]
HERMES MODEL
Hermes is a middleware for activity-based applications in distributed environments. The architecture of Hermes consists of three layers User Layer, System Layer, Run-Time Layer. User layer is based on workflows. User layer provides to users a set of programs for interacting with the workflow management system. System layer hosts Pool of Workflow Executors (WE). WEs are UserAgents generated from the User-Level Workflow (ULW) specification. Agent generation procedure is twofold 1) ULW is mapped to an Agent Level Workflow (ALW), 2) ALW is used to generate WEs. Run time layer provides a set of active services. Service Agents provide access to services.
Hermes Mobile Computing Platform consists of three layers Agent layer, Core layer, Basic Services layer. Agent Layer supports User agents and Service agents. The core layer supports identification, communication, loading and security. Basic Services layer supports discovery, mobility, creation, communication and security.[8]
Mole MODLE
Mole is a Java-based mobile agent system. It based on the concepts of agents and places. Places provide the environment and Agents are active entities. Agent identified by a globally unique agent identifier. Life cycle of agent consists of five process init, prepare, start, die, and migrate. Mole provide graphical agent monitor used for examine places, agents, and messages sent between the agents. Sand box security model is used.[9]
IV. Conclusion
Technique
Support Mobile Agent
Enhance security of agent
platform independent
Enhance Stability
MOLE
yes
yes
no
no
AGILLA
yes
no
no
yes
LIME ONE
yes
yes
no
no
HERMS
yes
yes
yes
no
M4WEB
GIS
yes
no
yes
yes
Soma
yes
yes
no
no
