Localisation of mobile devices has been a topic of academic research and industry practice for solving various application problems, examples can be footfall counting and profiling used for location based digital or physical advertising, crowd monitoring for public security, emergency handling, transport measurement and management, etc. Often the solutions for localisation require large scale networks hardware and/or software upgrades, which can be very costly. However, we note the fact that many commercial use cases actually do not require very high resolution of localisation and satisfactory level of accuracy may be sufficient for attaining business decision quality. A reasonable trade-off between achieving business value and minimising additional costs on network equipment purchase and maintenance is to build solutions that rely only on telco-network data and utilize data mining methods to improve the localisation accuracy of mobile devices that carried by subscribers. In this work, we aim to achieve acceptable accuracy for localisation at the resolution of region of interest (ROI), the exact shape of which is defined according to business requirements. One example is the geographical division of planning sub-zone in Singapore. We make use of the Global Positioning System (GPS) locations extracted from mobile broadband log that contain the longitude and latitude of the subscriber to annotate the telco-network data. We experimented with three learning models: maximum likelihood estimation, dominant serving ROI, and random forest, along with the baseline of localisation based on cellular tower locations. The experiment results demonstrate the effectiveness of the proposed models and demonstrate accuracy improvement from baseline of 37.8% (naive cellular tower localisation) to 78.4% (random forest classification).

This paper presents the ongoing work of applying the project-based learning approach to teaching IoT technologies in an advanced embedded networking course, which is offered to electrical and computer engineering students. The emphasis of this course is on teaching the prominent wireless networking technologies IEEE 802.11 (WiFi) for the IoT and the development of the IoT applications. Microcontrollers have been used to teach fieldbus networks or distributed real-time control systems. However, conventional efforts of teaching regular computer networks (e.g., the Ethernet) mostly make use of either personal computers (e.g., desktop, laptop) or single board computers installed with Linux or Windows operating system. The unique feature of this ongoing work is to use existing ARM-based microcontrollers without complex operating systems to teach wired and wireless network technologies in the IoT. Our primary experiences indicate that ARM-based microcontrollers with networking connectivity provide the flexible, cost-effective, well-supported platforms for developing the IoT applications.

Compared with conventional keyboard and mouse systems, touch sensing input devices provide more accurate and direct interaction with the user. More and more touch devices appear in various domains. To keep up with this technology change, we have introduced touch sensing technologies to electrical and computer engineering technology students by using the project-based learning (PBL) approach. Our experiences indicate that the PBL approach is efficient and practical for teaching touch sensing techniques. Two different sets of courseware including hardware kits and software packages have been utilized in design projects to teach touch in two existing courses respectively. Students have shown the great interest and the capability in adopting touch devices into their senior embedded systems design projects to improve user interactions with the computing systems.

Mutation testing suffers from the high computational cost of automated test-vector generation, due to the large number of mutants that can be derived from programs and the cost of generating test-cases in a white-box manner. We propose a novel algorithm for mutation-based test-case generation for Simulink models that combines white-box testing with formal concept analysis. By exploiting similarity measures on mutants, we are able to effectively generate small sets of short test-cases that achieve high coverage on a collection of Simulink models from the automotive domain. Experiments show that our algorithm performs significantly better than random testing or simpler mutation-testing approaches.

Bit-precise verification with variables modeled as bitvectors has recently drawn much interest. However, a huge search space usually results after bit-blasting. To accelerate the verification of bit-vector formulae, we propose an efficient algorithm to discover non-uniform encoding widths. Weof variables in the verification model, which may be smaller than their original modeling widths but sufficient to find a counterexample. Different from existing approaches, our algorithm is path-oriented, in that it takes advantage of the controllability and observability values in the structure of the model to guide the computation of the paths, their encoding widths and the effective adjustment of these widths in subsequent steps. For path selection, a subset of single-bit path-controlling variables is set to constant values. This can restrict the search from those paths deemed less favorable or have been checked in previous steps, thus simplifying the problem. Experiments show that our algorithm can significantly speed up the search by focusing first on those promising, easy paths for verifying those path-intensive models, with reduced, non-uniform bitwidth encoding.

We present a new verification approach that applies aggressive program slicing and a proof-based abstraction-refinement strategy to enhance the scalability of bounded model checking of embedded software. While many software model-checking tools use program slicing as a separate or optional step, our program slicing is integrated in the model construction and reduction process. And it is combined with the compilation optimization techniques so to compute a more accurate slice. We also explore a proof-based abstraction-refinement strategy using the under/overapproximation on our proposed software model, and propose a heuristic method of deciding new encoding size to refine the under-approximation. Experiments on C programs from wireless cognitive radio systems show this approach can greatly reduce the model size and shorten the solving time by the SAT-solver.