6
Cognitive Mesh Networks: Maintaining Transmission Quality Using Analysis and Prediction of Trafc with Articial Neural Networks Iuri Andreazza Unidade Acadêmica de Graduação UNISINOS São Leopoldo, Brazil [email protected] Dante Moreira Zaupa Unidade Acadêmica de Graduação UNISINOS São Leopoldo, Brazil [email protected] Mateus Raeder Unidade Acadêmica de Graduação UNISINOS São Leopoldo, Brazil [email protected]  Abstract—The need for greater quality of service demands more from the infrastructure of existing network models. Con- sidering the growth rate of devices on the network requesting data constantl y, these models will soon become obsol ete due to the ir inabil ity to pr ovide consta nt access to the net work without requiring the user to congure their devices each time the network changes. Mesh Network model provides constant access to the wide network, but it has several problems, such as the difculty of maintaining the quality of services by providing links with speed and perf orman ce desire d. In this paper we propose a new approach to package routing in the mesh model, in order to combine the exibility of Articial Neural Networks, aiming at mai ntaini ng net wor k per formance in the fac e of changing topology.  Keywords-Mesh Networks; Articial Neural Networks; Rout- ing Algorithms; Cognitive Routing; I. I NTRODUCTION At the end of the 20th cent ur y and be gi nning of the 21st century, the evolution of computing and communication technologies made mobile computing and wireless computer networks wides pread , which increase d their presence s in acade mic, industrial and home enviro nmen ts. This allows anyone to easil y access comp uter network s and data bases infor mation. The main issue that stress wire less networks is the inc reasing pre sen ce of not ebo oks , cel l pho nes and many other mobile devices that enjoy the infrastructure to access the Internet. With the availability of online services and tools for both desktop and mobile devices users, the need for constant information and connectivity becomes essential, as well as the quality of that information [1]. Data networks clients are more and more present in the digital world and they demand constant connectivity. With the advent of online and digital tools, the need for quality of connectio n ris es, for cing the cur ren t inf ras tru ctu re to almost ach ie ve its li mit . Alt hough net wor ks can sup por t a gre at amoun t of users wit h hig h le vel of qua lit y , these models did not predict new incoming users (common day- do-day devices) like refrigerators, cell phones, tablets, print- ers, information center, cars and houses, for example. The omnipresence of computational devices stresses the network structure, since it foresees that the whole environment will be con nec ted to the devi ces, thu s nee ding a con sta nt and intense data ow [2][1]. Most common wireless networks works in a centralized way, i.e., there is a single access point for all clients. In thes e scen arios, the network becomes expo sed to fail ures and if the access point becomes unavailable or unreachable, all of its clients lose the network connection. An existing pattern that provides a scenario in which even if the access point is down the connection among the clients still exists, is the mesh pattern. This model of network has some impor- tant functionalities such as self-repairing, self-conguration and self -lay out (amo ng other s), giv ing an adva ntage over other models [2]. However, the mesh model currently does not offer a goo d sup por t for quali ty of ser vic e, i.e., data transmission ins ide the mesh is slo w and ten ds to need retransmission, making it impractical to maintain a constant stream of data to the clients in the mesh. Wi th these new types of user s join ing the mesh infras - tructure out of research environments, this model becomes interesting for a large scale implementation. In this way, it is possible to make an environment in which new devices can  join the existing ones. This situation will naturally create a “computational ecosystem” of devices, besides reducing the need for users to manually congure each device. Howe ver, this growth of users insi de the mesh crea tes a tota lly stoc hast ic env ironment, which generat es a grea t los s of qua lit y in the rou ti ng pro ces s alg ori thms (in the mai nte nance of the data ow). This pro ble m is a ma in factor for the resistance against large scale adoption of the mes h model, not allowin g use rs to take adv ant age of its capabilities. In the con sta nt sea rch for mor e e xib le mec han isms, cap abl e of ada pti ng to the con sta nt cha nges in topolo gy (th at are natural in a mes h net wor k), we pro pos e the use of Articial Neural Networks (ANNs) to improve the mesh mode l pack et rout ing process. ANNs hav e the ability for non-linear and parallel processing, which gives the capacity to process and classify data in a cognitive way with high efciency. To make them efcient, it is necessary to train the ANNs

Cognitive Mesh Networks: Maintaining Transmission Quality Using Analysis and Prediction of Traffic with Artificial Neural Networks

Embed Size (px)

Citation preview

7/31/2019 Cognitive Mesh Networks: Maintaining Transmission Quality Using Analysis and Prediction of Traffic with Artificial N…

http://slidepdf.com/reader/full/cognitive-mesh-networks-maintaining-transmission-quality-using-analysis-and 1/6

7/31/2019 Cognitive Mesh Networks: Maintaining Transmission Quality Using Analysis and Prediction of Traffic with Artificial N…

http://slidepdf.com/reader/full/cognitive-mesh-networks-maintaining-transmission-quality-using-analysis-and 2/6

7/31/2019 Cognitive Mesh Networks: Maintaining Transmission Quality Using Analysis and Prediction of Traffic with Artificial N…

http://slidepdf.com/reader/full/cognitive-mesh-networks-maintaining-transmission-quality-using-analysis-and 3/6

7/31/2019 Cognitive Mesh Networks: Maintaining Transmission Quality Using Analysis and Prediction of Traffic with Artificial N…

http://slidepdf.com/reader/full/cognitive-mesh-networks-maintaining-transmission-quality-using-analysis-and 4/6

7/31/2019 Cognitive Mesh Networks: Maintaining Transmission Quality Using Analysis and Prediction of Traffic with Artificial N…

http://slidepdf.com/reader/full/cognitive-mesh-networks-maintaining-transmission-quality-using-analysis-and 5/6

7/31/2019 Cognitive Mesh Networks: Maintaining Transmission Quality Using Analysis and Prediction of Traffic with Artificial N…

http://slidepdf.com/reader/full/cognitive-mesh-networks-maintaining-transmission-quality-using-analysis-and 6/6