Cognitive Sensing in Smart Cities Using Optical Sensors

The concept of Smart Cities becomes paramount importance both to Government and Private Sectors. The increasing pressure of urbanization creates tremendous pressure to civic infrastructures like drainage, water system, electrical networks (Smart Grid), transportation etc [1] -- [3]. The use of Big Data Analytics is becoming a necessary tool to design and monitor Smart Cities. With the cost of sensors decreasing everyday, the flood of data is ubiquitous. We have designed a neural network based on cognitive algorithm for monitoring Smart Cities. Our algorithm is developed on the highly scalable distributed machine learning GraphLab Framework, embeds the deep learning features in the GraphLab Architecture to gain insight of the unstructured data through hierarchical abstraction. This is a scalable software architecture using Apache STORM. An Update Function (UF) is defined, which updates the scope of the data in the GraphLab. The Graph Rules, used by Graph Transformation, are defined to construct hidden layers in GraphLab to get more control over information flow. The "Update Function" and Graph Transformation trigger each other, which forms Hybrid GraphLab. The Hybrid GraphLab allows more flexible, fault tolerant, computational capabilities and possibilities to add more hidden multi-layers in tune with deep learning. Our algorithm can be extended to other applications like old age homes, hospitals etc for improved monitoring and management, with various sensors such as Gas Sensors, Light Sensors, Directionable Sensor Probes and Noise Sensors.