No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

This research program covers three topics relating to the humancomputer interface namely, voice recognition, tools and techniques for evaluation, and user and interface modelling. An investigation into the implementation of voice recognition technologies examines how voice recognisers may be evaluated in commercial software. A prototype system was developed with the collaboration of FEMVIEW Ltd. (marketing a CAD package). Proposals for future research using the prototype system suggests the need for field trials to assess its usefulness in a working environment and to gain insights to end-user attitudes. A new generation of voice system is proposed based around a phoneme-based pattern matching paradigm, natural language understanding facilities and intelligent knowledgebased systems capable of building on knowledge by inference and deduction. In order to assess the 'usability' of the FEMVIEW CAD software a subject-base formal evaluation was conducted which involved: - (1) the analysis of responses to a multi-user survey of end-user attitudes; (2) collecting behavioural performance measures from students learning to use the software. (3) cognitive and affective data obtained from laboratory experimentation using experienced users of the CAD package. A theoretical approach to evaluation leads to the hypothesis that human-computer interaction is affected by personality, influencing types of dialogue, preferred methods for providing help, etc. A user model based on personality traits, or habitual behaviour patterns (HBP) is presented. Proposals are given to use the HBP model in future self-adaptive interfaces. Results from experimentation to justify the model are inconclusive. Finally, a practical framework is provided for the evaluation of human-computer interfaces. It suggests that evaluation is an integral part of design and that the iterative use of evaluation techniques throughout the conceptualisation, design, implementation and postimplementation stages will ensure systems that satisfy the needs of the users and fulfil the goal of 'usability'. The major contributions made to the knowledge of this subject can be summarised as follows: (1) the practical problems of implementing voice recognition technologies in commercial software; (2) the development of a new personalised user model which accounts for individual's idiosyncrasies; (3) methods for applying simple evaluation techniques in order to assess software 'usability'; (4) a practical framework for developing usable software.

Dense deployment of small cells underlaying the traditional macrocells is considered as a key enabling technique for the emerging fifth generation (5G) cellular networks. However, the diverse transmit powers of the base stations (BSs) in such a network lead to uneven distribution of the traffic loads among different BSs when received signal power (RSP)-based user association is used. Moreover, provisioning of efficient and economical backhauling for these small cells is a crucial challenge. To combat this, wireless backhauling is been considered as a viable and cost-effective approach that allows operators to obtain end-to-end control of their network rather than leasing third party wired backhaul connections. But the scarcity of radio frequency (RF) spectrum in the licensed bands is still a major constraint which necessitates efficient spectrum planning for backhaul/access links of small cells. Emerging communications techniques such as full-duplexing, which allows transmission and reception in the same spectrum band, can be used to tackle the problem of spectrum scarcity. In the above contexts, the objective of the research work presented in this thesis is to develop efficient user association and wireless backhauling schemes for small cell networks and analyze their performances. In particular, i) A channel-access aware user association scheme is proposed to tackle the problem of uneven distribution of traffic load among different BSs, ii) Performance analysis of full-duplex (FD) wireless backhauling of small cells is carried out when compared to half-duplex (HD) wireless backhauling), iii) A method for downlink spectrum allocation for in-band and out-of-band wireless backhauling of full-duplex small cells is presented to optimally allocate spectrum for access and backhaul links, iv) A method for optimal channel and power allocation is presented for downlink access and backhaul links for half-duplex small cells. The proposed methods and performance analysis models will be useful for optimizing the design and deployment of small cell networks. / October 2017