Quality-aware Tasking in Mobile Opportunistic Networks - Distributed Information Retrieval and Processing utilizing Opportunistic Heterogeneous Resources.
Advances in wireless technology have facilitated direct communication among mobile devices in recent years, enabling opportunistic networks. Opportunistic networking among mobile devices is often utilized to offload and save cellular network traffic and to maintain communication in case of impaired...
| Summary: | Advances in wireless technology have facilitated direct communication among mobile devices in recent years, enabling opportunistic networks. Opportunistic networking among mobile devices is often utilized to offload and save cellular network traffic and to maintain communication in case of impaired communication infrastructure, such as in emergency situations. With a plethora of built-in capabilities, such as built-in sensors and the ability to perform even intensive operations, mobile devices
in such networks can be used to provide distributed applications for other devices upon opportunistic contact. However, ensuring quality requirements for such type of distributed applications is still challenging due to uncontrolled mobility and resource constraints of devices. Addressing this problem, in this thesis, we propose a tasking methodology, which allows for assigning tasks to capable mobile devices, considering quality requirements. To this end, we tackle two fundamental types of tasks required
in a distributed application, i.e., information retrieval and distributed processing.
Our first contribution is a decentralized tasking concept to obtain crowd collected data through built-in sensors of participating mobile devices. Based on the Named Data Networking paradigm, we propose a naming scheme to specify the quality requirements for crowd sensing tasks. With the proposed naming scheme, we design an adaptive self-organizing approach, in which the sensing tasks will be forwarded to the right devices, satisfying specified quality requirements for requested information.
In our second contribution, we develop a tasking model for distributed processing in opportunistic networks. We design a task-oriented message template, which enhances the definition of a complex processing task, which requires multiple processing stages to accomplish a predefined goal. Our tasking concept enables distributed coordination and an autonomous decision of participating device to counter uncertainty caused by the mobility of devices in the network. Based on this proposed model, we develop computation handover strategies among mobile devices for achieving quality requirements of the distributed processing.
Finally, as the third contribution and to enhance information retrieval, we integrate our proposed tasking concept for distributed processing into information retrieval. Thereby, the crowd-collected data can be processed by the devices during the forwarding process in the network. As a result, relevant information can be extracted from the crowd-collected data directly within the network without being offloaded to any remote computation entity. We show that the obtained information can be disseminated to the right information consumers, without over-utilizing the resource of participating devices in the network.
Overall, we demonstrate that our contributions comprise a tasking methodology for leveraging resources of participating devices to ensure quality requirement of applications built upon an opportunistic network. |
|---|