International Conference on Information and Communication Technology and Its Applications (ICTA 2016) Federal University of Technology, Minna, Nigeria November 28 – 30, 2016 A Review of Computer Network Reliability Godwill Udoh, Caroline Alenoghena, and Bala A. Salihu Department of Telecommunications Engineering, Federal University of Technology, Minna, Nigeria godwilludoh@gmail.com, carol@futminna.edu.ng, salbala2000@hotmail.com Abstract—Network reliability, as a field, is focused on There have been efforts by researchers using various questions of topological disconnection between data nodes. approaches in determining the reliability of a computer Computer network reliability problem is primarily resolved by network and as such are view to classify the approaches and calculating the probability that specific set of nodes in define most suitable method of analyzing understudied understudied network can speak with each other at a given Computer network is needed. time. Network reliability analysis cuts across the design, In this paper a review of computer network reliability deployment and testing of computer networks. Review reveal analysis is carried out and presented in five subheadings. two major approaches to reliability analysis of computer Section 2 gives the different approaches to reliability network these are Path/cute numeration and Case analysis analysis of computer networks, section 3 give an overview which utilizes the concept of graph decomposition. The measures of computer network reliability analysis include of related works, section 4 describe the model used, section issues on connectivity, capacity and travel time. The different 5 describes the measures of reliability analysis, section 6 parameters used in defining the reliability of computer- gives a description of different parameters that can affect communication system include; Failure Rate, Mean Time To network reliability, section 7 gives the evaluation of the Repair, Mean Time Between Failures, Mean Time To Failure reliability index and conclusion is presented in section 8. and Availability. Keywords-reliability analysis; computer network; reliability II. DIFFERENT APPROACHES TO RELIABILITY ANALYSIS index; network reliability OF COMPUTER NETWORKS Network Reliability defines the probability that an I. INTRODUCTION understudied Network can perform its stated function to an acceptable level of performance for some given period of Effective communications is critical in determining the time [1]. Any specific measure of reliability, thus, depends economic, social, innovative and military prowess of on the nature of the desired function and what levels of modern day society; this is hinged on the fact that effective performance are acceptable. Earliest works on Network communications is a key stone in all human activities. This reliability was around 1950, with connectivity of networks communication is majorly effected by data nodes. A data been used as there liability index [2]. node could be a mobile phone, information centers, Different set of algorithms are employed in reliability encrypting and decrypting machines which have evolved analysis of computer networks. The algorithms used can be into what we call computers in today‟s world. Thus, the grouped into two: design and deployment of reliable data nodes has been given a high priority and this has enhanced the study of Computer A. Path/Cut Enumeration: Network reliability analysis as a field on its own. Network reliability is concerned with the topological disconnection This entails the listing of all the simple paths between between data nodes. Computer network reliability problem end nodes. These represent a complete set of favorable is primarily resolved by calculating the probability that (unfavorable) non-disjoint events. Simple paths are links specific set of nodes in understudied network can speak in the network that connect set of nodes while prime cut sets with each other at a given time. Network reliability analysis are links in the network which when disconnected cause the cuts across the design, deployment and testing of computer network to fail. The simple paths are considered assets of networks. favorable events while the prime cuts as set of unfavorable As communication is now centered on computer events. Reliability analysis entails summing the terminal networks, the design of reliable computer networks is much reliabilities of these paths which is an indication that each needed. Reliability analysis of a computer-communication node communicates with a designated node. To obtain the network gives “worthiness test” of the infrastructure or computer network reliability, the inclusion-exclusion relevant components that constitute the computer network techniques of path and cuts is carried out. More efficient and as such, seeks to evaluate the relevance of the computer techniques based on Boolean algebra, can be utilized [2]– network to its intended design expectations. [7]. 198 International Conference on Information and Communication Technology and Its Applications (ICTA 2016) B. Case analysis: cater for issues of power supply, equipment malfunctioning Case analysis uses the method of graph decomposition. and working environment) in the reliability analysis of This entails the creation of subsets from the path sets, either computer networks around a reference edge or around a number of It is important to note that Failures in computer networks edges/links/paths. A reference edge is simply the node from cannot be completely eliminated and, hence, reliability can which the factoring is anchored. When more than one edge never be 100 percent. In order to reduce their negative is considered, graph decomposition is restricted to a impact on end-to-end connectivity, redundancy techniques conservative policy as against an exhaustive one. Using a are implemented on network routing so as to ensure a proper conservative policy minimizes the number of disjoint events alternate (backup) path is available upon failure of the in the analysis. Disjoint events are simple paths that are not primary working path. connected or have common node. This decomposition Performances of networks can be measured by simplifies the analysis and helps cancel out occurrence of comparing with the reliability index. The reliability index parallel links. represents the probability that a network operates. The performance of a design can be studied by changing the topology of the network. This is achieved by adding, removing or replacing some components so as to augment III. OVERVIEW OF RELATED WORKS the reliability or for satisfying certain specifications. Most of the works reviewed evaluate the reliability of understudied Networks by the methods of minpaths and mincuts. A minpath is the shortest distance/path/ number of hops between nodes needed to keep them up and IV. MODELS USED IN COMPUTER NETWORK RELIABILITY communicating while mincut is the smallest break in the link/network that renders the link/network ineffective. An Models used to describe computer networks help to aggregation of paths between nodes is called pathsets while define a frame in which the network could be studied. From an aggregation of cuts in the network is called cutsets. papers researched, the common model used for Computer As a network enlarges and nodes increase, the number of Network is the Stochastic Model. minpaths and mincuts increase exponentially. Effective A stochastic network [24] describes a physical system in analysis of these sets needed to keep the network up or down which each node and/or each edge (directed or undirected) via optimization methods help in estimating the reliability of can fail independently and failure one element does not understudied networks. affect another network element. A number representing the Genetic algorithm as an optimization tool was used in non-failure probability is assigned to the element. With this evaluating the reliability of networks [8]-[11]. model, the network reliability analysis problem is solely Wei-Chang Yeh [12] in his work used Particle Swarm calculating the probabilities of failure/operation for the edges Optimization and Monte Carlo Simulation in analyzing the and/or nodes needed to establish link connectivity [25]. pathsets in order to evaluate the reliability In the Reliability analysis of Computer Networks Monte Carlo simulation was also employed in [13] [14] (especially Ring Networks) [37], the following assumptions to analyze the pathsets/cutsets in order to evaluate the are made (depending on the number of components reliability understudied network. understudied): Wei Hou [15] in his thesis, analyzed the reliability of  Components are either operational or failed at any networks with software and hardware failures. He developed given time and Component state is a random event, models; MORIN – Modelling Reliability for Integrated s, independent of any other. Networks and SAMOT – Simplified Availability Modeling  The reliability of a network component is the Tool, with which he used in analyzing the Network. probability that it is operational at a given time. Boolean reduction technique was also used in evaluating  The Channel Capacity, C, is fixed and C » B (where the reliability of understudied network [16] – [21] B is the provisioned Bandwidth/Capacity for the Mathematical Analysis are also employed in the Network). reliability analysis of networks [22] – [23].  Failure of any electronic component in a station, An overview of works reviewed shows that little have including the power supply, causes the station to fail been done on reliability analysis of Ring topology networks  Network reliability does not include the probability (this might be a ripple-effect from the fact that most for failure of attached hosts since they are external to computer networks implement the mesh topology for its the understudied communications subnetwork. obvious advantages). Earlier works seem to use survivability, availability, susceptibility, connectivity and reliability interchangeably. The components used to define the node in the These mentioned parameters are distinct and a field of study understudied network are the link, port and station itself. The on themselves. station define the system health and caters for power issues Modern communication networks are made up of reliable in the node. Failure/success probabilities of these component components and failed components are quickly repaired. are independent of each other but have an overall effect on Multiple connections which allow for rerouting of traffic the state of the node. when network failure occurs is also a common feature. With The port comprises the transmitter, receiver and the the afore-stated in mind, it is important to take into account inbound/outbound link used in effecting self-healing when a the connection ports, links and state of the nodes (this should fault occurs. 199 International Conference on Information and Communication Technology and Its Applications (ICTA 2016) V. MEAURES OF COMPUTER NETWORK RELIABILITY  Failure Rate (FR) is the number of failures A good index for measuring the utilization of a computer experienced or expected for a device divided by the network reflects that a network fails gradually and that some total equipment operating time. nodes (links) are more important than others. The stated  Mean Time to Repair (MTTR) is the total time taken index is not also based on traffic patterns. Measures with the to effect corrective maintenances divided by the total above requirements include; terminal, “capacity-related”, number of these repairs. and “travel-time related” reliability measures.  Mean Time between Failures (MTBF) is the mean  time expected between failures, measured in hours. A. Terminal Reliability For constant failure rate systems, MTBF is the Terminal Reliability is the probability that there exists an inverse of the Failure Rate. end-to-end connection between two or more nodes in a  Equation 1givesthemathematical formula for computer network needed to keep the network up and evaluating MTBF running [9]. There are basically 3variants;K-Terminal,2- Terminal and All-Terminal Reliabilities. (1) 1) K-Terminal Reliability: The predominant measures of reliability when applied to computer networks are mainly  Mean Time to Failure (MTTF) is the mean time specialized cases of k-terminal reliability. This is defined as expected before the first failure of a piece of the probability that a path exists and connects k-terminals equipment. It is meant to be the mean over a long (nodes) within the network [26] -[28].This reliability is the period of time and a large number of units. sum of the probabilities of disjoint success paths [29].  Availability (A) is the probability of a system to 2) Two Terminal Reliability: This is the probability that being operational when called upon to perform its a path exists between end pair of nodes in the network [30]- function. Availability and unavailability (1-A) are [32]. often expressed as probabilities. 3) All Terminal Reliability: This is the probability that Equation 2 gives the mathematical formula for the nodes in a network are communicating with each other determining the availability of the network. [33][34]. Two terminal and all- terminal reliability are modifications on the K- Terminal measure where K=2 and K=n (n is the number of nodes in the network), respectively. (2) B. Capacity-Related Network Reliability VII. EVALUATION OF COMPUTER NETWORK RELIABILITY INDEX This field is an extension or specialization of the research in computer network reliability. It is acknowledge Reliability analysis of a computer network invariably that not all links (edges) may have the same capacity leads to evaluating a reliability index for the network. This (weight, bandwidth). Further, it is also considered that the is implied sincere liability, like availability, is expressed as capacity demanded of the network is variable. Capacity- probabilities. A reliability index defines a scale on which the Related computer network reliability defines the network reliability of a computer network can be measured. This reliability in terms of the bandwidth (variable or fixed) scale ranges from 0–1. A value of zero implies a failed needed to establish efficient data-throughput between nodes network while one means an “excellent” reliable network. [35] [36]. It places a bench mark requirement (bandwidth) The computer network reliability performance parameters needed for connectivity between node shaving fixed or described above are used in the evaluation of Reliability (R) variable link capacities. It is used when the data requirement index of the network. is same or more than the channel bandwidth and/or subscribed bandwidth. Computationally we have; C. Travel Time Network Reliability (3) This basically defines the time taken for data to travel between link ends or nodes. It also describes the time taken (4) for a hand shake to occur between specific nodes in a network and uses this as an index to measure reliability (‘T’ is the number of hours) VI. COMPUTER NETWORK RELIABILITY PERFORMANCE For multiple components; PARAMETERS Network reliability cannot be fully defined without (5) taking into consideration some parameters that affect computer network performance. Different parameters that („n‟ is the number of components in the system) affect network reliability of a computer-communication system are described below [37]. They include failure rate, 𝑛 𝑅(𝑇) = 𝑖=1 𝑅𝑖(𝑇) (6) Meantime to repair, mean time between failures, mean time to failure and availability. 200 International Conference on Information and Communication Technology and Its Applications (ICTA 2016) The reliability of a computer network is invariably a sum [10] K. Abdullah, and A. Smith. "A hybrid genetic algorithm approach for of the probabilities of the individual components of the backbone design of communication networks." Evolutionary Computation, 1999. CEC 99. Proceedings of the 1999 Congress on. communications aspect of the Computer communication Vol. 3. IEEE, 1999. network understudied (Equation 6). [11] Y. Wei-Chang. "An improved sum-of-disjoint-products technique for the symbolic network reliability analysis with known minimal VIII. CONCLUSION paths." Reliability Engineering & System Safety 92.2 (2007): 260- 268. The objective of a typical computer network reliability [12] W.Oliver "Terminal-pair reliability of three-type computer analysis is to obtain an index against which the reliability of communication networks." IEEE transactions on reliability 41.1 the understudied network can be measured. Though a (1992): 49-56. review of computer network reliability has been carried out [13] Niederreiter, Harald. Quasi‐Monte Carlo Methods. John Wiley & in this paper, further work is been done on implementing the Sons, Ltd, 2010. above stated discourse on an existing computer network. [14] H. Wei. Integrated Reliability and Availability Aanalysis of Networks The intent of this is to observe and document (if any) With Software Failures and Hardware Failures. Diss. University of South Florida, 2003. variations from the review, give recommendations for future expansion and deployment of network facilities and evaluate [15] R. Suresh, A. Kumar, and E. V. Prasad. "Computing terminal reliability of computer network." Reliability Engineering 16.2 (1986): the reliability of the understudied network. 109-119. This analysis can be carried out for a proposed network [16] S. Sieteng, and S. Rai. "CAREL: Computer aided reliability evaluator and an existing network. This is to ensure better choices of for distributed computing networks." IEEE Transactions on Parallel network components, network topology deployment and and Distributed Systems 2.2 (1991): 199-213. flexibility in expansion and upgrading of network. The [17] R. Suresh, and S. Sieteng "A computer approach for reliability Reliability analysis of Computer Networks forms a evaluation of telecommunication networks with heterogeneous link- veritable tool in defining the worth of a network to its capacities." IEEE Transactions on Reliability 40.4 (1991): 441-451. intended design expectations and as such seeks to give a [18] V. Malathi, and S. Trivedi. "An improved algorithm for symbolic reliability analysis." IEEE Transactions on Reliability 40.3 (1991): measure for user satisfaction. It invariably forms an 347-358. mportant criteria to enhancing Quality of Service– QoS. [19] K. Sy-Yen, F. Yeh, and H. Lin. "Efficient and exact reliability evaluation for networks with imperfect vertices." IEEE Transactions on Reliability 56.2 (2007): 288-300. ACKNOWLEDGMENT [20] B. Octavian. An Enhanced Approach to Network Reliability Using Boolean Algebra. Diss. Lafayette College, 2003. Mr. Godwill, U., Dr. Caroline, A., and Dr. Bala, S., thank the Department of Telecommunications in School of [21] O.C.Ibe, "Reliability comparison of token-ring network schemes." IEEE transactions on reliability 41.2 (1992): 288-293. Engineering of the Federal University of Technology, Minna for help given in this research work. [22] A. Satyanarayana and M. Chang. "Network reliability and the factoring theorem." Networks 13.1 (1983): 107-120. [23] G. Giancarlo. "Reliability evaluation of Common-Cause failures and REFERENCES other interdependencies in large reconfigurable networks." (2010). [24] G. Anna"A survey of statistical network models." Foundations and [1] M. Beccuti. "A new symbolic approach for network reliability Trends® in Machine Learning 2.2 (2010): 129-233. analysis." IEEE/IFIP International Conference on Dependable Systems and Networks (DSN 2012). IEEE, 2012. [25] H.Gary, C. Lucet, and N. Limnios. "K-terminal network reliability measures with binary decision diagrams." IEEE Transactions on [2] C. Cassir,and M. Bell. "Reliability of Transport Networks. Chapter 7. Reliability 56.3 (2007): 506-515. The N+ M Person Game Approach To Network Variability." Publication of: RESEARCH STUDIES PRESS [26] Y. Jiahnsheng, and C. B. Shin. "K-terminal reliability in ring LTD (2001). networks." IEEE Transactions on Reliability 43.3 (1994): 389-401. [3] Giaoutzi, Maria, and P. Nijkamp, eds. Network strategies in Europe: [27] S. Appajosyula, and R. Kevin Wood. "A linear-time algorithm for Developing the future for transport and ICT. Ashgate Publishing, computing K-terminal reliability in series-parallel networks." SIAM Ltd., 2008. Journal on Computing 14.4 (1985): 818-832. [4] Y. Wei-Chang. "A new Monte Carlo method for the network [28] R. Suresh, A. Kumar and E. V. Prasad. "Computing terminal reliability." Proceedings of First International Conference on reliability of computer network." Reliability Engineering 16.2 (1986): Information Technologies and Applications (ICITA2002). 2002. 109-119. [5] J. Wilson "An improved minimizing algorithm for sum of disjoint [29] J. Buzacott "A recursive algorithm for finding reliability measures products [reliability theory]." IEEE Transactions on Reliability 39.1 related to the connection of nodes in a graph." Networks 10.4 (1980): (1990): 42-45. 311-327. [6] S. Soh and S. Rai. "Experimental results on preprocessing of path/cut [30] Y. Oliver "Terminal-pair reliability of three-type computer terms in sim of disjoint products technique." IEEE transactions on communication networks." IEEE transactions on reliability 41.1 reliability 42.1 (1993): 24-33. (1992): 49-56. [7] K. Aggarwal, C. Chopra, and J. S. Bajwa. "Modification of cutsets for [31] D. Torrieri. "Calculation of node-pair reliability in large networks reliability evaluation of communication systems." Microelectronics with unreliable nodes." Microelectronics Reliability 3.36 (1996): 446. Reliability 22.3 (1982): 337-340. [32] C. Héctor, G. Rubino, and M. Urquhart. "An algorithm to compute [8] R. Dirk, F. Rothlauf and P. Gmilkowsky. "Designing reliable the all-terminal reliability measure." OPSEARCH-NEW DELHI- communication networks with a genetic algorithm using a repair 38.6/1 (2002): 567-579. heuristic." European Conference on Evolutionary Computation in [33] D. Berna, F. Altiparmak, and A. Smith. "Genetic algorithm design of Combinatorial Optimization. Springer Berlin Heidelberg, 2004. networks considering all-terminal reliability." Proceedings of the 6th [9] A. Fulya, B. Dengiz, and A. Smith. "Reliability optimization of Industrial Engineering Research Conference. 1997. computer communication networks using genetic [34] S. Ahuja, "Performance based reliability Optimization for computer algorithms." Systems, Man, and Cybernetics, 1998. 1998 IEEE networks." Southeastcon'97. Engineering new New Century., International Conference on. Vol. 5. IEEE, 1998. Proceedings. IEEE. IEEE, 1997. 201 International Conference on Information and Communication Technology and Its Applications (ICTA 2016) [35] L. Yi-Kuei. "Reliability of a stochastic-flow network with unreliable [37] P. Chinmayananda "Node Reliability in WDM Optical Network." branches & nodes, under budget constraints." IEEE Transactions on (2012). Reliability 53.3 (2004): 381-387. [36] R. Suresh, and S. Sieteng "A computer approach for reliability . evaluation of telecommunication networks with heterogeneous link- capacities." IEEE Transactions on Reliability 40.4 (1991): 441-451. 202