Abstract :
[en] This thesis studies the problem of energy efficiency of communications in distributed computing paradigms, including cloud computing, mobile cloud computing and fog/edge computing. Distributed computing paradigms have significantly changed the way of doing business. With cloud computing, companies and end users can access the vast majority services online through a virtualized environment in a pay-as-you-go basis. %Three are the main services typically consumed by cloud users are Infrastructure as a Service (IaaS), Platform as a Service (PaaS) and Software as a Service (SaaS).
Mobile cloud and fog/edge computing are the natural extension of the cloud computing paradigm for mobile and Internet of Things (IoT) devices. Based on offloading, the process of outsourcing computing tasks from mobile devices to the cloud, mobile cloud and fog/edge computing paradigms have become popular techniques to augment the capabilities of the mobile devices and to reduce their battery drain. Being equipped with a number of sensors, the proliferation of mobile and IoT devices has given rise to a new cloud-based paradigm for collecting data, which is called mobile crowdsensing as for proper operation it requires a large number of participants.
A plethora of communication technologies is applicable to distributing computing paradigms. For example, cloud data centers typically implement wired technologies while mobile cloud and fog/edge environments exploit wireless technologies such as 3G/4G, WiFi and Bluetooth. Communication technologies directly impact the performance and the energy drain of the system. This Ph.D. thesis analyzes from a global perspective the efficiency in using energy of communications systems in distributed computing paradigms. In particular, the following contributions are proposed:
- A new framework of performance metrics for communication systems of cloud computing data centers. The proposed framework allows a fine-grain analysis and comparison of communication systems, processes, and protocols, defining their influence on the performance of cloud applications.
- A novel model for the problem of computation offloading, which describes the workflow of mobile applications through a new Directed Acyclic Graph (DAG) technique. This methodology is suitable for IoT devices working in fog computing environments and was used to design an Android application, called TreeGlass, which performs recognition of trees using Google Glass. TreeGlass is evaluated experimentally in different offloading scenarios by measuring battery drain and time of execution as key performance indicators.
- In mobile crowdsensing systems, novel performance metrics and a new framework for data acquisition, which exploits a new policy for user recruitment. Performance of the framework are validated through CrowdSenSim, which is a new simulator designed for mobile crowdsensing activities in large scale urban scenarios.
