A multi-modal smart sensing network for marine environmental monitoring
Zhang, Dian (2015) A multi-modal smart sensing network for marine environmental monitoring. PhD thesis, Dublin City University.
Full text available as:
There is an imperative need for long-term, large-scale marine monitoring systems that will allow decisions to be made based on the analysis of collected data to avoid or limit negative impacts on the ecosystem. Modern marine environmental sensing technologies, such as autonomous wireless sensor networks (WSNs), provide the capability to meet the challenges of high spatial and temporal scales. However, the significant amount of data generated from WSNs is a significant challenge for manual analysis. These multitudinous data need to be automatically processed, indexed and catalogued in a smarter way that can be more easily understood, accessed and managed by operators, scientists and policy makers. Moreover, current research works show that WSNs have their own limitations, for example, reliability issues and the fact that they are passive systems and provide context-less data. Thus, it is becoming increasingly clear that in order to adequately monitor marine environments, they need to be characterised from multiple perspectives. Combining multiple technologies and sensing modalities in environmental monitoring programmes can provide not only advantages of reliability and robustness for sensing systems, but also enhanced understanding of environmental processes. In addition, considerable advances can be made if robust sensing technology can be combined with sophisticated methods of data analysis, classification and cataloguing. The aim of this work is to bridge the gap between current aquatic monitoring systems and futuristic ideal large scale multi-modality smart sensing networks for marine environmental monitoring. To illustrate this, a smart sensing system is proposed and two case studies are used to show data processing from in-situ measurements and from camera based visual sensing data automatically using machine learning techniques. Abnormal events detection results from an in-situ sensor and shipping traffic detection results from visual sensor are combined to illustrate the benefit of coupling multiple sensing modalities.
Archive Staff Only: edit this record