Proaktiivse mobiilsusteadliku uduandmetöötluse suunas
Abstract
Paljude värkvõrk- ja ärirakenduste tavapäraseks osaks on sõltuvus kaugete pilveteenuste poolt pakutavast andmetöötlusvõimekusest. Arvestatav hulk seesugustest rakendustest koguvad andmeid mitmetelt ümbritsevatelt heterogeensetelt seadmetelt, et pakkuda reaalajal põhinevaid teenuseid oma kasutajatele. Taolise lahenduse negatiivseks küljeks on aga kõrge viiteaeg, mis muutub eriti problemaatiliseks, kui vastava rakenduse efektiivne töö on väleda vastuse saamisega otseses sõltuvuses. Taolise olukorra puhul on viiteaja vähendamiseks välja pakutud uduandmetöötlusel põhinev arhitektuur, mis kujutab endast arvutusmahukate andmetöötlusühikute jaotamist andmeallikate ja lõppkasutajatele lähedal asuvatele arvutusseadmetele. Vaatamata sellele, et uduandmetöötlusel põhinev arhitektuur on paljutõotav, toob see kaasa uusi väljakutseid seoses kvaliteetse uduandmetöötlusteenuse pakkumisega mobiilsetele kasutajatele. Käesolev magistritöö käsitleb proaktiivset lähenemist uduandmetöötlusele, kasutades selleks lähedalasuvatel kasutajatel baseeruvat mobiilset ad hoc võrgustikku, mis võimaldab uduteenusetuvastust ja juurdepääsu ilma pilveteenuse abi kasutamata. Proaktiivset lähenemist kasutatakse nii teenusetuvastuse ja arvutuse migratsiooni kui ka otsese uduteenuse pakkumise käigus, kiirendades arvutusühikute jaotusprotsessi ning parendadades arvutuste jaotust vastavalt käitusaegsele kontekstiinfole (nt. arvutusseadmete hetkevõimekus). Lisaks uuriti uduarvutuse rakendusviisi mobiilses sotsiaal–silmusvõrgustikus, tehes andmeedastuseks optimaalseima valiku vastavalt kuluefektiivsuse indeksile. Lähtudes katsetest nii päris seadmete kui simulaatoritega, viidi läbi käesoleva magistritöö komponentide kontseptuaalsete prototüüpide testhindamine.
A common approach for many Internet of Things (IoT) and business applications is to rely on distant Cloud services for the processing of data. Several of these applications collect data from a multitude of proximity-based ubiquitous resources to provide various real-time services for their users. However, this has the downside of resulting in explicit latency of the result, being especially problematic when the application requires a rapid response in the edge network. Therefore, researchers have proposed the Fog computing architecture that distributes the computational data processing tasks to the edge network nodes located in the vicinity of the data sources and end-users, to reduce the latency. Although the Fog computing architecture is promising, it still faces challenges in many areas, especially when dealing with support for mobile users. Utilizing Fog for real-time mobile applications faces the new challenge of ensuring the seamless accessibility of Fog services on the move. Further, Fog computing also faces a challenge in mobility when the tasks originate from mobile ubiquitous applications in which the data sources are moving objects. In this thesis, a proactive approach for Fog computing is proposed, which supports proactive Fog service discovery and process migration using Mobile Ad hoc Social Network in proximity, enabling Fog-assisted ubiquitous service provisioning in proximity without distant Cloud services. Moreover, a proactive approach is also applied for the Fog service provisioning itself, in order to hasten the task distribution process in Mobile Fog use cases and provide an optimization scheme based on runtime context information. In addition, a case study regarding the usage of Fog Computing for the enhancement of Mobile Mesh Social Network was presented, along with a resource-aware Cost-Performance Index scheme to assist choosing the approach to be used for transmission of data. The proposed elements have been evaluated by utilizing a combination of real devices and simulators in order to provide proof-of-concept.
A common approach for many Internet of Things (IoT) and business applications is to rely on distant Cloud services for the processing of data. Several of these applications collect data from a multitude of proximity-based ubiquitous resources to provide various real-time services for their users. However, this has the downside of resulting in explicit latency of the result, being especially problematic when the application requires a rapid response in the edge network. Therefore, researchers have proposed the Fog computing architecture that distributes the computational data processing tasks to the edge network nodes located in the vicinity of the data sources and end-users, to reduce the latency. Although the Fog computing architecture is promising, it still faces challenges in many areas, especially when dealing with support for mobile users. Utilizing Fog for real-time mobile applications faces the new challenge of ensuring the seamless accessibility of Fog services on the move. Further, Fog computing also faces a challenge in mobility when the tasks originate from mobile ubiquitous applications in which the data sources are moving objects. In this thesis, a proactive approach for Fog computing is proposed, which supports proactive Fog service discovery and process migration using Mobile Ad hoc Social Network in proximity, enabling Fog-assisted ubiquitous service provisioning in proximity without distant Cloud services. Moreover, a proactive approach is also applied for the Fog service provisioning itself, in order to hasten the task distribution process in Mobile Fog use cases and provide an optimization scheme based on runtime context information. In addition, a case study regarding the usage of Fog Computing for the enhancement of Mobile Mesh Social Network was presented, along with a resource-aware Cost-Performance Index scheme to assist choosing the approach to be used for transmission of data. The proposed elements have been evaluated by utilizing a combination of real devices and simulators in order to provide proof-of-concept.