Monte Carlo simulation-based algorithms for estimating the reliability of mobile agent-based systems
Introduction
Mobile agents have received much attention in the last decade as a new distributed programming paradigm. An agent in a mobile agent-based system (MABS) has the ability to migrate to distributed recourses via the network, to perform the required operations locally, and therefore eliminating the network transfer of intermediate data (Moizumi, 1998; Brewington et al., 1999). Performance issues of MABS have not been studied, especially at the level of performance modeling. The main reason behind this is the complexity of the MABS characteristics. This complexity may jeopardize the validity of modeling, if it is not discovered and analyzed in an efficient way.
The reliability of the computer communication network (CCN) is a factor that may affect the performance and availability of MABS (Hsieh and Hsieh, 2003; Murphy and Picco, 2002). Additionally, CCN reliability may affect agent's strategy as well. In this paper, which builds on our previous work in Daoud and Mahmoud (2005b), we define the MABS as a system that consists of a number of independent agents, each of which accomplishes an independent task. We study the reliability of a MABS with respect to the network status and its conditions. We propose a random agent planning strategy with static property to randomize the experiment. The static property means that we create a route to mobile agents before sending them out, and the route is not updated dynamically as agents roam the network (Baek et al., 2002).
The rest of this paper is organized as follows. Section 2 presents the problem statement and discusses the modeling of CCNs and MABS. The proposed algorithms and their details are presented in Section 3. Section 4 presents simulation results and discusses them. Section 5 discusses the validity of the proposed model. Related works are presented in Section 6. Finally, the conclusion and future work are discussed in Section 7.
Section snippets
Problem statement
A mobile agent network system is composed of two main components or subsystems, the CCN and the MABS that uses this network (Lovrek and Sinkovic, 2001) as shown in Fig. 1. Our goal in this paper is to estimate the reliability of MABS with respect to network status. Reliability estimation can be evaluated with respect to different assumptions. To be able to predict reliability, we must make concrete assumptions about the system under study. As a first step toward our goal, we need to model the
Proposed algorithms
In this section we propose three algorithms to estimate RS. Each algorithm deals with a different case. In the first algorithm, the reliability of a MABS is estimated by taking the long-term average of mean number of mobile agents in the system that successfully complete their tours (or walks) with respect to status of CCN. We use Monte Carlo simulation to estimate the required reliability as given in Cancela and El Khadiri (1995).
Experimental results and discussion
Using the network depicted in Fig. 2, we investigate the reliability of the MABS for k=1, 2, 3, 4, 5. Fig. 3a shows the simulation results of R_MABS1 algorithm when pi=0.9 for all i and qj=1 for all j.
It is clear that the reliability of the MABS vary between 0.727 and 0.737 despite the number of agents. From this result, we see that the difference between the worst case and best case is 0.010, which is a small value. Therefore, this result demonstrates the robustness of the estimation method
Validity of the model
The open question here is this: does the proposed simulation model purely reflect the behavior of the real system or the conditions under investigation of the real system? In fact, we are seeking for the validity of the simulation algorithms and their results.
Without loss of generality, we now discuss the validity issue from conceptual point of view. First, the level of abstraction plays a key role in building any model, and precisely, any model cannot cover all the system details. Thus, the
Related work
Reliability analysis of MABS is a complex issue for which little attention has been paid. Most of the work done in this area is related to distributed systems. A number of algorithms have been proposed to estimate the reliability of distributed systems and distributed applications.
Raghavendra et al. (1988) introduced two reliability measures: distributed program reliability and distributed system reliability. An efficient approach based on graph traversal is developed to evaluate the proposed
Conclusion and future work
In this paper, we have presented the fundamental definitions and theories for estimating the reliability of mobile agent-based systems (MABS) based on network conditions. We developed two algorithms using the Monte Carlo technique to estimate the required reliability. We assumed that the mobile agents are independent but operate simultaneously. The experimental results demonstrate the robustness of the proposed algorithms and the estimation method used. The reliability of the underlying network
Acknowledgments
The authors would like to thank the anonymous reviewers for the many helpful suggestions for improving this paper. This work was supported in part by the Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant No. 045635.
References (16)
- et al.
Reliability and cost optimization in distributed computing systems
Comput Oper Res
(2003) - et al.
Estimation of all-terminal network reliability using an artificial neural network
Comput Oper Res
(2002) - Altiparkmak F, Dengiz B, Smith AE. Reliability optimization of computer communication networks using genetic...
- Baek J, Yeo J, Yeom H. Agent chaining: an approach to dynamic mobile agent planning. In: Proceedings of the IEEE 22nd...
- et al.
Mobile agents in distributed information retrieval
- et al.
A recursive variance reduction algorithm for estimating-network reliability
IEEE Trans Reliab
(1995) - et al.
Reliability analysis of distributed systems based on a fast reliability algorithm
J IEEE Trans Parallel Distributed Systems
(1992) - Daoud M, Mahmoud QH. Reliability analysis of mobile agent-based systems. In: Proceedings of the 20th annual ACM...
Cited by (17)
A new method for reliability allocation of avionics connected via an airborne network
2015, Journal of Network and Computer ApplicationsCitation Excerpt :Among these assumptions, the first two are frequently used in reliability allocation problems, e.g., Coit and Smith (1996), Kuo and Zuo (2002), Kroese et al. (2007) and Khalili-Damghani et al. (2013). However, multi-state systems and common-cause failures may exist in regular engineering systems, which have been further studied by Tian et al. (2008) and Daoud and Mahmoud (2008), respectively, but not for networked systems. In this paper, we only focus on two-state systems with independent failures.
ANALYZING FACTORS OF GUI SIMULATION AS LEARNING MEDIA TOWARD STUDENTS’ LEARNING OUTCOMES
2022, Journal of Technology and Science EducationA review of agent-based modeling in the cooperative sector of economics
2021, Studies in Systems, Decision and ControlDesigning secure and reliable mobile agent based system for reliable MANET
2019, International Journal of Ad Hoc and Ubiquitous ComputingA methodology for ranking of MAS reliability measures
2017, Proceeding - IEEE International Conference on Computing, Communication and Automation, ICCCA 2016Kinematic reliability solution of 3-UPS-PU parallel mechanism based on monte carlo simulation
2015, Open Mechanical Engineering Journal