The last few years have seen an exceptional growth in the number and use of personal mobile devices, including smart phones and tablets. This growth has been followed by the diffusion of smart objects and sensors, as well as high-resolution displays, not only at home, but also in workplaces and even in public spaces such as malls, theaters, and concert halls. It is straightforward to interact with mobile devices in different ways, as they are equipped with a touch screen, a microphone, one or more cameras, and several additional sensors. Unfortunately, screens and embedded sensors are mostly output devices; as a consequence, they are not very friendly when it comes to user interactions. In almost all cases, they lack input capabilities at all. These constraints call for different methods of interaction and novel techniques for configuration and management.
My research on pervasive displays has taken an approach based on its lifecycle: a newly installed cloud-based screen has to be initially configured (i.e., bootstrapped), then remotely managed and maintained. I have developed a system for secure bootstrapping of pervasive displays that leverages mobile devices and existing enterprise wireless technologies (best paper at the UbiComp 2014 conference). I have then characterized the requirements for remote management of pervasive display deployments and evaluated different solutions for replicating their content (e.g., to build management dashboards). Finally, I have explored applications for interacting with pervasive displays through mobile devices.
My research on the Internet of Things has focused on its heterogeneity and the resource constraints of the participating devices. I have designed a storage infrastructure suitable for heterogeneous and multimedia data in the Internet of Things. I have also evaluated different databases for efficient data storage in the same scenario. I have finally established the feasibility in using the security infrastructure of mobile network operators to authenticate resource-constrained devices with no input capabilities (best paper at the IoT 2015 conference).
Despite the ever increasing storage and processing capabilities of mobile devices, computing on-the-move still relies on the availability of remote resources. These resources can be provided, for instance, by a cloud computing infrastructure. In one of the most extreme cases, mobile devices simply act as thin clients: they send input events to a remote server and receive the updated content of the screen. Such a thin-client computing paradigm is a form of mobile cloud computing. In this context, I have evaluated how different systems and protocols affect the performance and the user experience of running remote applications in the cloud. I have specifically considered two representative use cases: corporate applications (i.e., office productivity suites) and cloud gaming.
Wireless communications enable effective mobile device networking. However, they are affected by several issues including reliability, spectrum availability, and power consumption. Reliability is indeed a critical factor, as many important applications (e.g., those involving data collection in a wireless sensor network) cannot work without source data. To this end, I have investigated how taking inspiration from nature can help improve the reliability of wireless sensor networks. I have also addressed lifetime maximization through load balancing and latency-sensitive data collection in the same reference scenario.
Wireless spectrum is a scarse resource. Mobile network operators are particularly facing the shortage of radio channels to serve an ever increasing number of customers. In this context, I have investigated how to leverage unused TV whitespace to offload video traffic from cellular networks. I have also realized a system for bidirectional communications between smartphones using visible light.