Agent Approaches to Role-Based SecurityAgent Approaches to Enforce Role-Based Security inDistributed and Web-Based Computing*S. A. Demurjian, Sr., Y. He, T. C. Ting, and M. SabaDepartment of Computer Science and EngineeringThe University of Connecticut191 Auditorium Road, U-155Storrs, CT 06269, USAEmail: {steve,ting,saba}@engr.uconn.eduTel. 860-486-3719 * The work in this paper has been partially supported by a contract from the Mitre Corp. (Eatontown, NJ) and a researchgrant by AFOSR.AbstractIn the age of information technology, organizations of alltypes are seeking to effectively utilize and disseminateinformation, by designing and developing dependable andsecure distributed computing environments that allowexisting and future systems to inter-operate. While manyexisting access control approaches (mandatory,discretionary, and role-based) can be leveraged for thesupport of security in distributed and web-based settings,their assumptions of a centralized computing model may beinsufficient in a distributed setting. In recent years, agentcomputing has emerged as a new computing paradigm,particularly suited to distributed and web-basedapplications. This paper explores software agents, focusingon their ability to support role-based security in a dynamic,object-based setting which is suitable for distributed andweb-based applications. The agent approaches differ intheir utilization of agents (stationary and mobile) and thegranularity level of the involved classes/objects. We alsoreport on our prototyping efforts using aglets, a Java-basedmobile agent model from IBM.1. IntroductionToday's and tomorrow's distributed and web-basedapplications will be comprised of existing legacy,commercial-off-the-shelf (COTS), and databaseapplications that interact with new clients, servers,and web-information repositories, as organizationsstrive to allow information to be utilized in new andinnovative ways. The fundamental challenge facingorganizations, software architects, system designers,and application builders (i.e., stakeholders) involvesthe ability to leverage computing and networkingresources, and data and information in existingapplications, to construct new dependable and securedistributed and web-based applications. Our focus ison the secure nature of distributed and web-basedapplications, particularly in venues which promoteelectronic banking, commerce, informationdissemination (push and pull), and so on. Distributedand web-based applications will require researchersand practitioners to design security solutions thatexpand and transcend available alternatives.Traditional alternatives such as mandatory accesscontrol (MAC) [Keef88, Land84], discretionaryaccess control (DAC) [Loch88, Sand96], and role-based security (RBS) [Demu97] may all be useful tosome degree as stakeholders and security engineerswork towards the establishment of a cohesive securitypolicy. Agent computing, which first emerged justfive years ago [Gene94], has great potential forsupporting the design/development of securedistributed and web-based applications.Agents act on the behalf of individuals (users),in individual or collaborative fashion, to assist in aparticular task, hopefully making it easier for users toaccomplish what they intended. Software agents, ascomputing objects have a specific function orresponsibility to perform, and can be defined inAgent Approaches to Role-Based Securityformal terms to have a state and a behavior within theruntime environment. Software agents have fourmandatory properties [Lang98]: ability to sense andreact to changes in the environment; autonomouscontrol over its own state and behavior; proactive toachieve specific goals (typically of the user); and,constantly executing within the runtime environment.Stationary agents are restricted to a single computingnode in accomplishing their tasks. Mobile agents canmigrate to a new location in order to execute theirrequired responsibilities. All agents are like otherobjects in that they can be created and destroyed.However, agents cannot interact by invoking eachothers methods; rather, they communicate viamessage passing.The autonomous and mobile nature of agentsmake them attractive for security purposes. Forexample, one scenario could have agents dynamicallycreated from a client application to carry out thesecure access to objects across a network. In such ascenario, the agent may go forth and visit multiplenodes to collect/update relevant objects.Alternatively, multiple mobile agents may besimultaneously dispatched, each processing a singlerequest, to achieve parallel processing in a distributedsetting. Security can be included as part of the agentfunctionality, providing a specificity of the role of theclient that dispatched the agent. Note also thatmobile agents are a significant security concern froman execution perspective, due to their potential abilityto act as a threat, and that there must also beprotection of the agent's state from a malicious host.These two critical issues are beyond the scope of thispaper.Our intent in this paper is to present and explorevarious scenarios for agent approaches to securitywithin a role-based context for distributed and web-based applications. We view this as a crucial andimportant first step in understanding the differentways that agent computing models can be utilized tosupport role-based security. Moreover, if agentcomputing continues to increase in popularity andusage, it is incumbent upon the research communityto provide security solutions. While the technology isdangerous from a security perspective, it is criticalthat we offer strong and proven solutions to guaranteevarying levels of secure access in a distributed settingthat utilizes agents. The research community mustreact to rapidly changing and newly emergingtechnologies.The work presented in this paper is a progressionof our own efforts [Demu97, Demu98, Smar99], withthe justification to pursue this effort influenced byother researchers [Hale96, Hale98, Tari98]. In ourprevious work [Demu98], we explored softwarearchitectures alternatives that utilized a client/serverparadigm to explore role-based security. While therewas distribution in some of the alternatives, theemphasis was on relatively static architectures, whichis inadequate for true dynamic, distributed and web-based applications. There have been efforts toinvestigate security for distributed objects [Hale96],which has recently been extended to provide a
View Full Document
Unlocking...