| dc.description.abstract | The research progress of communication has brought a lot of novel technologies to meet the multi-dimensional demands such as pervasive connection, low delay and high bandwidth. Device-to-Device (D2D) communication is a way to no longer treat the User Equipment (UEs) as a terminal, but rather as a part of the network for service provisioning. This thesis decouples UEs into service providers (helpers) and service requesters. By collaboration among proximal devices, with the coordination of cellular networks, some local tasks can be achieved, such as coverage extension, computation o oading, mobile crowdsourcing and mobile crowdsensing. This thesis proposes a generic framework Proximity as a Service (PaaS) for increasing the coverage with demands of service continuity. As one of the use cases, the optimal helper selection algorithm of PaaS for increasing the service coverage with demands of service continuity is called ContAct based Proximity (CAP). Mainly, fruitful contact information (e.g., contact duration, frequency, and interval) is captured, and is used to handle ubiquitous proximal services through the optimal selection of helpers. The nature of PaaS is evaluated under the Helsinki city scenario, with movement model of Points Of Interest (POI) and with critical factors in uencing the service demands (e.g., success ratio, disruption duration and frequency). Simulation results show the advantage of CAP, in both success ratio and continuity of the service (outputs). Based on this perspective, metrics such as service success ratio and continuity as a service evaluation of the PaaS are evaluated using the statistical theory of the Design Of Experiments (DOE). DOE is used as there are many dimensions to the state space (access tolerance, selected helper number, helper access limit, and transmit range) that can in uence the results. A key contribution of this work is that it brings rigorous statistical experiment design methods into the research into mobile computing. Results further reveal the influence of four factors (inputs), e.g., service tolerance, number of helpers allocated, the number of concurrent devices supported by each helper and transmit range. Based on this perspective, metrics such as service success ratio and continuity are evaluated using DOE. The results show that transmit range is the most dominant factor. The number of selected helpers is the second most dominant factor. Since di erent factors have di erent regression levels, a uni ed 4 level full factorial experiment and a cubic multiple regression analysis have been carried out. All the interactions and the corresponding coe cients have been found. This work is the rst one to evaluate LTE-Direct and WiFi-Direct in an opportunistic proximity service. The contribution of the results for industry is to guide how many users need to cooperate to enable mobile computing and for academia. This reveals the facts that: 1, in some cases, the improvement of spectrum e ciency brought by D2D is not important; 2, nodal density and the resources used in D2D air-interfaces are important in the eld of mobile computing. This work built a methodology to study the D2D networks with a di erent perspective (PaaS). | en_US |
