Depression does not only affect the personal life of individuals and their families and social circles but it has also a strongly negative economic impact as shown in several reports. According to a recent study, workers in the United Kingdom suffer high levels of depression than those anywhere else in Europe. The survey found that 1 in 10 employees had taken time off at some point in their working lives because of depression problems. Novel strategies for tackling the problem of depression and preventing suicides are needed. We believe that new emerging technologies, in particular mobile ones, together with the possibility of mining large amount of data in real-time can help to tackle this problem in new and more effective ways. Existing interview-based studies have shown that depression is significantly associated with a marked decline of physical activity. The goal of this project is to investigate how mobile phones can be used to collect and analyse mobility patterns of individuals in order to understand how mental health problems affect their daily routines and behaviour and how potential changes can be automatically detected. In particular, mobility patterns and levels of activity can be quantitatively measured by means of mobile phones, exploiting the GPS receiver and the accelerometers embedded in the devices. The data can be extremely helpful to understand the behaviour of a depressed person, and in particular, to detect potential changes in his or her behaviour, which might be linked to a worsening depressive state. By monitoring this information in real-time, health officers and charity workers might intervene by means of digital behaviour intervention delivered through mobile phones or by means of traditional methods such as by inviting the person for a meeting or by calling him or her by phone. In order to support these novel applications, it is necessary to build mathematical tools for analysing the mobility traces in real-time for the detection of gradual or sudden changes related to the emotional states of the individual. More specifically, we plan to devise analytical techniques for studying the relationships between human mobility patterns and emotional states. We plan to use existing datasets of human mobility and to collect data by means of a smartphone application distributed to people affected by depression. This can be considered as a sort of pilot study for a wider deployment of these technologies and it will provide a sound theoretical basis for further studies in this area. Finally, a key aspect of the proposed research work is the implementation of mechanisms for preserving the privacy of the individuals involved in the study.

In this project, entitled FlexCap, we propose to study the use of Morello's hardware capabilities to provide safety and isolation in Operating Systems (OSes). We propose to enable capability support and evaluate its efficiency in two OSes: FlexOS, a library OS offering a highly-configurable isolation profile that can be customised at build time towards specific use cases; as well as Unikraft, a high-performance/low latency unikernel. FlexOS allows the user to specialise the isolation/safety strategy of the operating system seamlessly at build time. Several fundamental parameters are customisable, including the granularity of kernel components isolation as well as the hardware mechanism used to enforce that isolation. FlexOS currently supports the Intel Memory Protection Keys and Extended Page Table mechanisms. Porting the OS to Morello would enable to benefit from the efficient compartmentalization brought by capabilities. In particular, the fine-grained memory protection and high degree of scalability resulting from the use of hardware capabilities should increase performance as well as security security, and decrease memory footprint in FlexOS, compared to the other mechanisms currently supported by the OS. FlexOS is itself an extension of the Unikraft unikernel, so porting FlexOS to Morello will first require porting Unikraft to the platform. Unikraft is a high-performance/low-latency unikernel targeting cloud applications. The high degree of performance it provides is achieved by running the application and the kernel code inside a single, completely unprotected address space. This obviously raises security concerns and porting Unikraft to Morello will allow to explore bringing back safety into high-performance unikernels, leveraging the security benefits brought by the capabilities' ability to provide safe versions of legacy programming languages (i.e. pure/hybrid capabilities). Finally, we also propose to explore advanced use of capabilities in FlexOS and Unikraft by studying 1) the possibility of incremental porting of Unikraft to pure capabilities and 2) horizontal compartmentalization of FlexOS components. These two OSes, FlexOS and Unikraft, are unique use cases for the application of Morello's hardware capabilities, and have never been explored in this context. These use cases differ significantly from the two OSes already available in the CHERI/Morello's software ecosystem (CheriBSD and CheriOS). CheriBSD is a general purpose monolithic OS and is unlikely to achieve the high level of performance of Unikraft. It also does not lend itself to flexible isolation like FlexOS. Further, CheriOS is unlikely to offer the same performance and compatibility with existing applications as Unikraft/FlexOS.

With the advent of the Internet of Things (IoT), machine type communications (MTC), cloud computing and many other applications, the wireless network will become far more complex, while at the same time far more essential than ever before. Given the above exponential growth in both connectivity and complexity of the wireless systems and the unprecedented demands on latency, capacity, ultra-reliability and security, the network is becoming analytically intractable. Naturally, human-driven physical layer (PHY) design approaches rooted on mathematical models of communications systems and networks which drive today's network architectures are being surmounted by the sheer complexity of the emerging network paradigms. Hardware imperfections, that are inevitable with the employment of low-cost MTC sensors and transmitters, will drastically increase the volatility of the network, and theoretically driven solutions typically relying on generic and highly inaccurate models cannot address this as they are highly sub-optimal in practice. The above challenges necessitate new data-driven approaches to the design of communications systems, as opposed to traditional system-model driven designs that are becoming obsolete. Towards the diverse communication paradigms of MTC of the future, there is an urgent need to address reliable and adaptive links detached from mathematical models, and instead based on data-driven approaches. This visionary project will address these fundamental challenges by developing new Neural Netowrk architectures tailored for wireless communications, and new transceiver architectures based on data-driven training. Our research will address the development of a) a communications specific DL framework, b) DL-inspired PHY solutions and, c) proof-of-concept verification of the proposed solutions. LeanCom will be performed with Huawei, NEC Europe, Duke University, The Digital Catapult and CommNet and aspires to kick-start an innovative ecosystem for high-impact players among the infrastructure and service providers of ICT to develop and commercialize a new generation of learning-based networks. The implementation, experimentation and testing (within WP3) of the proposed solutions serves as a platform towards commercialisation of the results of LeanCom, aiming towards an impact of a foundational nature for the UK's digital economy.

Mobile phone users are expected to exceed 5 billion in 2010 and the use of online social networks is soaring (Facebook alone has more than 500 million users). Today's mobile phones represent a powerful computing platform, given their ability to sense through a variety of sensors (e.g. accelerometer, Bluetooth, microphone, and magnetometer), their processing and communication capabilities. Phones are part of everyday life, and therefore represent an exceptionally suitable tool for investigating behaviour and promoting behaviour change, while social networks provide a valuable source of data about user preferences and social interactions. This proposal will investigate the power and challenges of using mobile phones for behaviour change interventions. This will involve tackling the challenges of measuring many aspects of human behaviour through power-limited mobile phones as well as integrating the information extracted through the phones with social data gathered on online social networks.Digital Behaviour Change Interventions (DBCIs) are interactive, automated packages of advice and ongoing support for behaviour change, which typically include: personalised advice based on responses to questions assessing needs, circumstances and preferences; support for goal-setting, planning and progress monitoring; automated reminders and progress-relevant feedback and encouragement; access to social support by email, online forums etc. DBCIs can be used for a wide range of different behaviours; for example, to reduce risky or antisocial behaviour, increase productivity in the workplace, enhance learning activities, or support environmentally important lifestyle change, such as reducing energy use. DBCIs are a relatively new method of supporting behaviour change, as the technology to support this kind of personalised interactive support is only now becoming available. They provide scientists with a means of carrying out detailed assessments of the process of behaviour change from a much larger sample of the population than has previously been possible.Traditional DBCIs have mainly been delivered by PCs and provide feedback to users based on their answers to questions about their activities and feelings. Our aim is to use mobile phone technology and online social networking applications to gather this kind of information during daily life without the need for users to answer questions. Mobile phones can be employed to sense whether the user is active, their mood, and who they are with or talking to, while online social networks can provide information about users' attitudes and social contacts. This information can then be used to deliver exactly the right kind of messages to users at the right time, depending on what the user is doing and feeling.We will work closely with users to develop ethical, acceptable and practical methods of measurement and behavioural intervention. We will then demonstrate and experimentally test the capabilities, performance and effectiveness of our tools and techniques by developing a range of DBCIs to address a major public health problem, weight management. We will recruit very large samples of people to try these DBCIs from our 'MyPersonality' population of 3 million Facebook users who have previously taken part in our studies. We will develop new methods to analyse the information we gather across time and space from a very large number of people. We aim to develop an in-depth understanding of how and why different people react to and use different intervention components. This will help us, and others, to design more popular and helpful DBCIs in the future.The tools we develop will be designed to be easily reusable and adaptable by others for different types of behaviour change. To ensure that a wide multidisciplinary community can benefit from the tools and methods for behaviour change that we develop we will provide extensive online educational and training materials, workshops, and exchanges.

AI applications have become pervasive: from mobile phones and home appliances to stock markets, autonomous cars, robots and drones. As AI takes over a wider range of tasks, we gradually approach the times when security laws, or policies, ultimately akin to Isaac Asimov's "3 laws of robotics" will need to be established for all working AI systems. A homonym of Asimov's first name, the project AISEC (``Artificial Intelligence Secure and Explainable by Construction"), aims to build a sustainable, general purpose, and multidomain methodology and development environment for policy-to-property secure and explainable by construction development of complex AI systems. We will create and deploy a novel framework for documenting, implementing and developing policies for complex deep learning systems by using types as a unifying language to embed security and safety contracts directly into programs that implement AI. The project will produce a development tool AISEC with infrastructure (user interface, verifier, compiler) to cater for different domain experts: from lawyers working with security experts to verification experts and system engineers designing complex AI systems. AISEC will be built, tested and used in collaboration with industrial partners in two key AI application areas: autonomous vehicles and natural language interfaces. AISEC will catalyse a step change from pervasive use of deep learning in AI to pervasive use of methods for deep understanding of intended policies and latent properties of complex AI systems, and deep verification of such systems.