16282

14531

كارشناسي ارشد

مخابرات (شبكه هاي مخابراتي)

اصفهان : دانشگاه صنعتي اصفهان

1399

هشت, 74ص.: مصور, جدول, نمودار

حسين سعيدي

علي قياسيان

فرامرز هندسي, مسعود رضا هاشمي

1399/11/18

كتابنامه

برق

مهندسي برق و كامپيوتر

1399/11/18

2671707

An SDN based Algorithm for Caching Routing and Load balancing in ICN Mohamad bagher Tavasoli mb tavasoli@gmail com October 2020 Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc Language Farsi Supervisor Dr Hosein Saidi hsaidi@iut ac ir Advisor Dr Ali Ghiasian ghiasian ali@sku ac ir Abstract Since a decade ago there has been a steady increase now in the deployment and the universal utilization of web basedsocial media such as Video on demands online games and social networks in a way that the mass data produced by themnow have a significant share of the whole Internet traffic Users are more interested in content than its location Thisissue led to a new concept for data management in networking called ICN which in contrast to host to host paradigms ofnetwork communication like IP networks communications are based on the publishing and the subscribing of the mostpopular network contents But ICN could not be implemented due to the profound incompatibility it has with the current IP network On the otherhand Software Define Network SDN has been introduced with an approach of separating the control and the data planewhich has led to significant improvements in the flexibility and the management of networks and has made the deploymentof ICNs feasible The initial idea is to combine the two ICN and SDN technologies to create a content oriented internet tohave effective control and management over the networks However this new approach itself gives rise to a new challengethat is the identification of the most appropriate places for data storage in the topology of the network In this research we investigate the management of contents and traffic in ICN using an SDN s Central Controller with the goal of servingthe requests by the nearest servers and simultaneously distributing and balancing traffic among all links and routers sothe congestions are prevented For this purpose we first formulate our model as a mixed integer non linear programmingproblem Then we try to find the solution in an Offline mode in which all the requests are known In advance we useGAMS commercial software with a random topology and we compare the outcomes with the ones of IP networks In thisscenario the results are developed in a way that the contents are saved on the nearest server or the available servers and themost optimized routes from the edge node to the servers are constructed In this case the results show that the proposedmodel consumes up to 20 less power than traditional networks and increases network efficiency by up to 50 Next we consider a dynamic scenario however since MINLP problems are categorized as NP Hard problems and finding anoptimal solution in GAMS are either time consuming or infeasible we have proposed a time scaled approximate algorithmwhich reduces the computational complexities At last to demonstrate the capabilities and performance of this algorithm we compare our results to the results of the GAMS and we show that our algorithm has extensively less computational time while the outcomes accurately resemble the outcomes of the GAMS as a semi optimal solution

حسين سعيدي

علي قياسيان

فرامرز هندسي, مسعود رضا هاشمي