The topic of this thesis is Opportunistic Networks (OPNETS), a type of mobile ad hoc network in which data are propagated by the movement of the network devices and by short-range wireless transmissions. This allows data to spread to many devices across large distances without the use of any infrastructure or powerful hardware. OPNET technology is in its fairly early stages of development and has a lot of potential for research. There are many applications that could benefit from OPNETS, such as sensor networks or social networks. However, before the technology can be used with confidence, research must be undertaken to better understand its behaviour and how it can be improved. In this thesis, the way in which information propagates in an OPNET is studied. Methodical parameter studies are performed to measure the rate at which information reaches new recipients, the speed at which information travels across space, and the persistence of information in the network. The key parameters being studied are device density, device speed, wireless signal radius and message transmission time. Furthermore, device interaction schemes based on epidemiological models are studied to find how they affect network performance. Another contribution of this thesis is the development of theoretical models for message spread in regions of one-dimensional (1D) and two-dimensional (2D) space. These models are based on preliminary theoretical models of network device interaction; specifically, the rate at which devices move within range of each other and the length of time that they remain within range. A key contribution of this thesis is in acknowledging that data transmissions between devices do not occur instantaneously. Due to latency in wireless communications, the time taken to transmit data is proportional to the amount of data being transferred. Non-instantaneous transmissions may fail before completion. Investigation is made into the effect this has on the rate of information propagation in OPNETS.

With network enabled consumer devices becoming increasingly popular, the number of connected devices and available services is growing considerably - with the number of connected devices es- timated to surpass 15 billion devices by 2015. In this increasingly large and dynamic environment it is important that users have a comprehensive, yet efficient, mechanism to discover services. Many existing wide-area service discovery mechanisms are centralised and do not scale to large numbers of users. Additionally, centralised services suffer from issues such as a single point of failure, high maintenance costs, and difficulty of management. As such, this Thesis seeks a Peer to Peer (P2P) approach. Distributed Hash Tables (DHTs) are well known for their high scalability, financially low barrier of entry, and ability to self manage. They can be used to provide not just a platform on which peers can offer and consume services, but also as a means for users to discover such services. Traditionally DHTs provide a distributed key-value store, with no search functionality. In recent years many P2P systems have been proposed providing support for a sub-set of complex query types, such as keyword search, range queries, and semantic search. This Thesis presents a novel algorithm for performing any type of complex query, from keyword search, to complex regular expressions, to full-text search, over any structured P2P overlay. This is achieved by efficiently broadcasting the search query, allowing each peer to process the query locally, and then efficiently routing responses back to the originating peer. Through experimentation, this technique is shown to be successful when the network is stable, however performance degrades under high levels of network churn. To address the issue of network churn, this Thesis proposes a number of enhancements which can be made to existing P2P overlays in order to improve the performance of both the existing DHT and the proposed algorithm. Through two case studies these enhancements are shown to improve not only the performance of the proposed algorithm under churn, but also the performance of traditional lookup operations in these networks.

An information technology that can combine the physical world and virtual world is desired. The Internet of Things (IoT) is a concept system that uses Radio Frequency Identification (RFID), WSN and barcode scanners to sense and to detect physical objects and events. This information is shared with people on the Internet. With the announcement of the Smarter Planet concept by IBM, the problem of how to share this data was raised. However, the original design of WSN aims to provide environment monitoring and control within a small scale local network. It cannot meet the demands of the IoT because there is a lack of multi-connection functionality with other WSNs and upper level applications. As various standards of WSNs provide information for different purposes, a hybrid system that gives a complete answer by combining all of them could be promising for future IoT applications. This thesis is on the subject of `Federated Sensor Network' design and architectural development for the Internet of Things. A Federated Sensor Network (FSN) is a system that integrates WSNs and the Internet. Currently, methods of integrating WSNs and the Internet can follow one of three main directions: a Front-End Proxy solution, a Gateway solution or a TCP/IP Overlay solution. Architectures based on the ideas from all three directions are presented in this thesis; this forms a comprehensive body of research on possible Federated Sensor Network architecture designs. In addition, a fully compatible technology for the sensor network application, namely the Sensor Model Language (SensorML), has been reviewed and embedded into our FSN systems. The IoT as a new concept is also comprehensively described and the major technical issues discussed. Finally, a case study of the IoT in logistic management for emergency response is given. Proposed FSN architectures based on the Gateway solution are demonstrated through hardware implementation and lab tests. A demonstration of the 6LoWPAN enabled federated sensor network based on the TCP/IP Overlay solution presents a good result for the iNET localization and tracking project. All the tests of the designs have verified feasibility and achieve the target of the IoT concept.

This research proposes a new architecture for Adaptive Educational Hypermedia Systems (AEHS). Architectures in the context of this thesis refer to the components of the system and their communications and interactions. The architecture addresses the limitations of AEHS regarding interoperability, reusability, openness, flexibility, and limited tools for collaborative and social learning. It presents an integrated adaptive and collaborative Web-based learning environment. The new e-learning environment is implemented as a set of independent Web services within a service-oriented architecture (SOA). Moreover, it uses a modern Learning Management System (LMS) as the delivery service and the user interface for this environment. This is a two-way solution, whereby adaptive learning is introduced via a widely adopted LMS, and the LMS itself is enriched with an external - yet integrated - adaptation layer. To test the relevance of the new architecture, practical experiments were undertaken. The interoperability, reusability and openness test revealed that the user could easily switch between various LMS to access the personalised lessons. In addition, the system was tested by students at the University of Nottingham as a revision guide to a Software Engineering module. This test showed that the system was robust; it automatically handled a large number of students and produced the desired adaptive content. However, regarding the use of the collaborative learning tools, the test showed low levels of such usage.

Rowing is an intensive, all-body sport, where bad technique can lead to injury. Crew cohesion, particularly timing, is vital to the performance of the boat. The coaching process, and injury prevention, could be enhanced if data relating to the movement of the oarsmen could be collected, without hindrance to the oarsmen, during on-water training. Literature until recently has concentrated upon boat-centric measurement. Advances in wireless technology have made feasible the collection of data from multiple physically separate sites, including on-body. After analysis of candidate radio standards, a Zigbee wireless Body Sensor Network (BSN) was designed and developed to synchronously collect data from several sensors across the wireless BSN. By synchronising sensor nodes via scheduled synchronising messages from the central coordinating node, synchronisation within 0.79msec ±0.39ms was achieved. Minimisation of the on-time of the sensor node radios currently extends the battery life by a factor of 5. Acceleration and muscle activity data collected using the wireless BSN was compared to data synchronously collected using proven motion analysis techniques to validate the system. Synchronous muscle activity data was collected via the wireless BSN from several muscles during both land-based and on-water rowing and the results compared. The system was proven to facilitate the identification of bad rowing technique, as well as differences in muscle recruitment between land- and water-based rowing. Data collection from a rowing crew was also demonstrated, and their muscle activity and inter-crew timing analysed. With an additional sensor node upon the boat, it is possible to correlate acceleration and muscle activity from the oarsman with acceleration of the boat itself. A novel, power-optimised wireless sensor network has been designed and demonstrated to facilitate on-water rowing monitoring that can be extended beyond single oarsman measurements to analyse the interaction and cohesion of a crew and their impact upon boat performance.

Ubiquitous computing (ubicomp) involves systems that attempt to fit in with users’ context and interaction. Researchers agree that system adaptation is a key issue in ubicomp because it can be hard to predict changes in contexts, needs and uses. Even with the best planning, it is impossible to foresee all uses of software at the design stage. In order for software to continue to be helpful and appropriate it should, ideally, be as dynamic as the environment in which it operates. Changes in user requirements, contexts of use and system resources mean software should also adapt to better support these changes. An area in which adaptation is clearly lacking is in ubicomp systems, especially those designed for mobile devices. By improving techniques and infrastructure to support adaptation it is possible for ubicomp systems to not only sense and adapt to the environments they are running in, but also retrieve and install new functionality so as to better support the dynamic context and needs of users in such environments. Dynamic adaptation of software refers to the act of changing the structure of some part of a software system as it executes, without stopping or restarting it. One of the core goals of this thesis is to discover if such adaptation is feasible, useful and appropriate in the mobile environment, and how designers can create more adaptive and flexible ubicomp systems and associated user experiences. Through a detailed study of existing literature and experience of several early systems, this thesis presents design issues and requirements for adaptive ubicomp systems. This thesis presents the Domino framework, and demonstrates that a mobile collaborative software adaptation framework is achievable. This system can recommend future adaptations based on a history of use. The framework demonstrates that wireless network connections between mobile devices can be used to transport usage logs and software components, with such connections made either in chance encounters or in designed multi–user interactions. Another aim of the thesis is to discover if users can comprehend and smoothly interact with systems that are adapting. To evaluate Domino, a multiplayer game called Castles has been developed, in which game buildings are in fact software modules that are recommended and transferred between players. This evaluation showed that people are comfortable receiving semi–automated software recommendations; these complement traditional recommendation methods such as word of mouth and online forums, with the system’s support freeing users to discuss more in–depth aspects of the system, such as tactics and strategies for use, rather than forcing them to discover, acquire and integrate software by themselves.

Detecting and dealing with congestion in delay tolerant networks is an important and challenging problem. Current DTN forwarding algorithms typically direct traffic towards particular nodes in order to maximise delivery ratios and minimise delays, but as traffic demands increase these nodes may become unusable. This thesis proposes Café, an adaptive congestion aware framework that reduces traffic entering congesting network regions by using alternative paths and dynamically adjusting sending rates, and CafRep, a replication scheme that considers the level of congestion and the forwarding utility of an encounter when dynamically deciding the number of message copies to forward. Our framework is a fully distributed, localised, adaptive algorithm that evaluates a contact’s next-hop potential by means of a utility comparison of a number of congestion signals, in addition to that contact’s forwarding utility, both from a local and regional perspective. We extensively evaluate our work using two different applications and three real connectivity traces showing that, independent of the network inter-connectivity and mobility patterns, our framework outperforms a number of major DTN routing protocols. Our results show that both Café and CafRep consistently outperform the state-of-the-art algorithms, in the face of increasing traffic demands. Additionally, with fewer replicated messages, our framework increases success ratio and the number of delivered packets, and reduces the message delay and the number of dropped packets, while keeping node buffer availability high and congesting at a substantially lower rate, demonstrating our framework’s more efficient use of network resources.

Wireless sensor networks (WSNs) have advanced rapidly in recent years and the volume of raw data received at an endpoint can be huge. We believe that the use of local knowledge, acquired from sources such as the surrounding environment, users and previously sensed data, can improve the efficiency of a WSN and automate the classification of sensed data. We define local knowledge as knowledge about an area that has been gained through experience or experimentation. With this in mind, we have developed a three-tiered architecture for WSNs that uses differing knowledge-processing capabilities at each tier, called the Knowledge-based Hierarchical Architecture for Sensing (KHAS). A novel aligning ontology has been created to support K-HAS, joining widely used, domain-specific ontologies from the sensing and observation domains. We have shown that, as knowledge-processing capabilities are pushed further out into the network, the profit - defined as the value of sensed data - is increased; where the profit is defined as the value of the sensed data received by the end user. Collaborating with Cardiff University School of Biosciences, we have deployed a variation of K-HAS in the Malaysian rainforest to capture images of endangered wildlife, as well as to automate the collection and classification of these images. Technological limitations prevented a complete implementation of K-HAS and an amalgamation of tiers was made to create the Local knowledge Ontology-based Remote-sensing Informatics System (LORIS). A two week deployment in Malaysia suggested that the architecture was viable and that, even using local knowledge at the endpoint of a WSN, improved the efficiency of the network. A simulation was implemented to model K-HAS and this indicated that the network became more efficient as knowledge was pushed further out towards the edge, by allowing nodes to prioritise sensed data based on inferences about its content.

This thesis consists of three independent and self-contained chapters regarding information and networks. The abstract of each chapter is given below. CHAPTER 1: The seminal “agreeing to disagree” result of Aumann (1976) was generalized from a probabilistic setting to general decision functions over partitional information structures by Bacharach (1985). This was done by isolating the relevant properties of conditional probabilities that drive the original result – namely, the “Sure-Thing Principle” and “like-mindedness” – and imposing them as conditions on the decision functions of agents. Moses & Nachum (1990) identified conceptual flaws in the framework of Bacharach (1985), showing that his conditions require agents’ decision functions to be defined over events that are informationally meaningless for the agents. In this paper, we prove a new agreement theorem in information structures that contain “counterfactual” states, and where decision functions are defined, inter-alia, over the beliefs that agents hold at such states. We show that in this new framework, decisions are defined only over information that is meaningful for the agents. Furthermore, the version of the Sure-Thing Principle presented here, which accounts for beliefs at counterfactual states, sits well with the intuition of the original version proposed by Savage (1972). The paper also includes an additional self-contained appendix in which our framework is re-expressed syntactically, which allows us to provide further insights. CHAPTER 2: We develop a parsimonious and tractable dynamic social network formation model in which agents interact in overlapping social groups. The model allows us to analyze network properties and homophily patterns simultaneously. We derive closed-form analytical expressions for the distributions of degree and, importantly, of homophily indices, using mean-field approximations. We test the comparative static predictions of our model using a large dataset from Facebook covering student friendship networks in ten American colleges in 2005, and we calibrate the analytical solutions to these networks. We find good empirical support for our predictions. Furthermore, at the best-fitting parameters values, the homophily patterns, degree distribution, and individual clustering coefficients resulting from the simulations of our model fit well with the data. Our best-fitting parameter values indicate how American college students allocate their time across various activities when socializing. CHAPTER 3: We examine three models on graphs – an information transmission mechanism, a process of friendship formation, and a model of puzzle solving – in which the evolution of the process is conditioned on the multiple edge types of the graph. For example, in the model of information transmission, a node considers information to be reliable, and therefore transmits it to its neighbors, if and only if the same message was received on two distinct communication channels. For each model, we algorithmically characterize the set of all graphs that “solve” the model (in which, in finite time, all the nodes receive the message reliably, all potentially close friendships are realized, and the puzzle is completely solved). Furthermore, we establish results relating those sets of graphs to each other.

We present an approach to the problem of learning by observation in spatially-situated tasks, whereby an agent learns to imitate the behaviour of an observed expert, with no direct interaction and limited observations. The form of knowledge representation used for these observations is crucial, and we apply Qualitative Spatial-Relational representations to compress continuous, metric state-spaces into symbolic states to maximise the generalisability of learned models and minimise knowledge engineering. Our system self-configures these representations of the world to discover configurations of features most relevant to the task, and thus build good predictive models. We then show how these models can be employed by situated agents to control their behaviour, closing the loop from observation to practical implementation. We evaluate our approach in the simulated RoboCup Soccer domain and the Real-Time Strategy game Starcraft, and successfully demonstrate how a system using our approach closely mimics the behaviour of both synthetic (AI controlled) players, and also human-controlled players through observation. We further evaluate our work in Reinforcement Learning tasks in these domains, and show that our approach improves the speed at which such models can be learned.