Periodic Reporting for period 1 - SmartPosition (Smartphone-based Mobile Positioning System)
Reporting period: 2016-05-02 to 2017-11-01
Given the challenges (i.e. harsh, unknown and dynamic environments) and opportunities (highly informative sensor data), this project has targeted to two tasks:
T1. develop a new concept, “Big Sensor”, as the definition of the collective employment of multi-type sensors in a single, centralized or distributed, system e.g. a smartphone; and
T2. establish new LPT solutions that complements the failure of Bayesian filters for the Big Sensor system, including two paradigms: Clustering for Filtering (C4F), and Fitting for Smoothing (F4S).
Based on the innovative integration of data mining (clustering and fitting) technologies with statistical signal processing, the project was carried out precisely to the scheduled objectives in our proposal, leading to three directions of the achievements of the Project:
Objective O1. Theoretically analyse when and why the filters/smoother are not workable or are less favorable than the sensor data-only inference, providing systematic guidelines to distinguish them;
Objective O2. Establish Big Sensor-oriented C4F and F4S algorithms and technologies, addressing their advantages and challenges regarding sensor data correlation, inconsistency, ambiguity, etc.
Objective O3. Develop a smartphone-based mobile positioning application, called “SmartPosition”, with the dual intention of verifying the theoretical findings and seeking realistic application for Home Care.
WP 1: General Coordination and Career Development (Period: 2016.5-8)
Works: definition of the interfaces; recognition of the needs of the project; literature review.
WP 2: Problem Formulation and Challenge Analysis, to achieve objective O1. (2016.6-10)
Works: effectiveness analysis of Bayesian filters for O1, problem formulation and challenge analysis regarding the use of a smartphone. Relevant variables and conditions identification.
WP 3: Research (I): Clustering for Filtering (C4F) (2016.9-2017.2)
Works: establishing efficient multi-sensor data clustering algorithms for target detection and estimation, in particular with respect to sensors embedded in a smartphone; a half part of objective O2.
Review of WP 4: Research (II): Fitting for Smoothing (F4S) (2017.1-2017.7)
Works: Exploring F4S theories to obtain continuous-time track estimation; the second half part of objective O2.
Review of WP 5: Development: Application for Home Care system (2017.7-2017.11)
Works: Integrating the ‘SmartPosition’ app to the Home Care system that the BISITE group has developed.
Review of WP 6: Dissemination, Exploitation, Public Engagement, Knowledge Transfer and Risk Management (2016.5-2017.11)
Works: Dissemination, Exploitation, Public Engagement, Knowledge Transfer and Risk Management have been carefully and maximally exploited over the whole process of the project.
The mains results are presented in the 10+ high-impact scientific papers and an open access mobile positioning APP. We have managed individual websites for real-time reporting the outcomes of the Project.
An mobile positioning APP open access at: https://bisite.usal.es/en/apps/smart-position
And simulation source codes related to the majority of the above publications: https://sites.google.com/site/tianchengli85/matlab-codes
1) “Big Sensor” concept and Joint smoothing, tracking and forecasting (STP) framework
“Big sensor” has identified a research opportunity to develop new LPT theories and algorithms, such as allowing for joint STP. With the rapid development of hardware sensors, the Big Sensor concept and the joint STP framework have expected to inspire more data-driven solutions and to promote the development of cutting-edge LPT solutions.
2) C4F/F4S paradigms
Both data clustering and numerical fitting are originally applied in the context of target detection and state estimation in cluttered environments, contributing to both theoretical advancements and practical implementations of state-of-the-art mobile positioning.
3) Effectiveness definition of the general Bayesian filter/smoother
The sensor-oriented C4F and F4S take a more practical approach by setting a worse bound on the mean error of any “effective” estimator, which addresses a fundamental issue by providing a clear and theoretical-solid definition of the effectiveness of the general Bayesian filter/smoother, offering a solid guideline for filter/smoother evaluation and design.
II. Expected results until the end of the project
Scientific outcomes of the Project are mainly reflected in 10+ high-quality scientific papers. The fruitful and enjoyable collaboration between the Host and Partner groups, based on their complementary nature, has advanced the European science excellence and made very positive impacts on their development (to be explained below). Their collaboration network has been enhanced. In addition, the Project has investigated excellent resources and experiences for communication and public engagement of the research activities, receiving high visibility both from the Public and the Academic.
III. Potential impacts
1) The Host and the Partner groups have made an enormous effort to design and to provide effective and specific training and programme courses to the researcher, according to the project needs, by which the Researcher has greatly broadened and diversified the competences of the Researcher in terms of new expertise, knowledge and skills (software + hardware; theory+ programming).
2) The project has enhanced the expertise and competence of the Host groups, strengthening its scientific and technological bases. In particular, the creative outcome of the project has been directly transferred to contribute to related projects that are being investigated in the Host group such as the following Regional and EU projects
HIGIA: Intelligent Platform for the Management and Tracking of Healthcare Personnel and Patients in Hospital Environments, Funding body: Castilla y León Regional Government, 2013-2017.
My-TRAC: My TRAvel Companion, Funding body: H2020 programme Shift2Rail, http://www.my-trac.eu/
3) Based on the promising potential value of the implemented mobile positioning APP, the Researcher and the Supervisors are keen in converting the final outcome of this project into socio-economic benefits for the EU and seek further involvement in new European projects that are related to Home Care, LPT and EGNSS/Galileo service, such as integrating the yielded technologies and protocols in the e-Bikemotion (www.ebikemotion.com) project that the Host group is continuously investigating and that has received market efficiency already.