SS49: Fog Computing, Deep Learning and Big Data Analytics

Scope of the Session: The huge data flow anticipated on the Internet due to the large number of IOT devices and sensors expected to be connected to the Internet can overload and slow down the Internet. This can also result in network congestion and make it infeasible to deploy any latency sensitive applications that require response in real time such as emergency health management and traffic management, as they cannot deliver the expected results in real time. This is due to the fact that the data generated by the IOT devices at the edge is required to traverse from the edge or remote field location like an agricultural field or a hospital to the cloud servers located at the large data centres , usually situated at long distances from the edge IOT devices and sensors. Thus, it will become infeasible to deploy cloud based IOT applications which are latency sensitive. Hence, a new paradigm of Fog Computing has been proposed where in the Fog server is located at a close proximity to the edge IOT devices and sensors which generate the data. No latency problems will be encountered, as there is no Internet connection required for them. The IOT devices and sensors are directly connected to the Fog server which can independently process the data it receives and respond back with results in real time, thereby overcoming all the latency related problems and issues in application deployment. The interaction of Fog server with its own Cloud server is of very limited nature, thereby unaffected by the problems of network latency or large data transmission.
In this workshop, the focus is on the research challenges that originate in the context of Fog Computing and Fog ecosystem: Fog architectures, Fog applications, including but not limited to smart cities, smart villages, smart health care, smart vehicle and vehicular Fog computing, augmented reality applications, Fog based latency aware application management and distributed application development, resource coordination, optimisation and migration methodologies, Fog security techniques and Fog Privacy issues.
Mining and extracting meaningful patterns from massive input data for decision making, prediction and inferencing is at the core of Big Data AnalytIcs.The additional challenges of Fog computing ecosystem , calling for new or extended machine learning and Big Data Analytics algorithms for fast moving stream data analytics,high dimensionality,scalability of algorithms,imbalanced input data,unsupervised and uncategorised input data,limited supervised /labelled data are all research challenges to be addressed.Adequate data storage,data indexing/tagging,fast information retrieval are the other key problems that are required to be addressed.Consequently ,innovative data management solutions are required to be deployed when working with Big Data in the context of Fog Computing.
Deep Learning techniques are relevant and applicable in the context of Big Data Analytics in the above stated Fog computing ecosystem,as they have superior quality of feature identification and selection for the input raw data.Superior quality of results have been acquired in Computer Vision and Speech Recognition for video and audio data respectively.By such application of Deep Learning techniques, it is possible to identify high level,complex abstractions from large or very large sizes of raw, unsupervised data.In the context of convergence of Deep Learning and Big Data Analytics for Fog computing, the functions such as Fast Information Retrieval,Semantic Indexing,DataTagging can be handled by Deep Learning techniques on to domain specific raw and unsupervised data for superior quality results.The non linear relationships and complex patterns in data in Big Data Scenarios cannot be identified by shallow Machine Learning algorithms because they are not as efficient as Deep Learning algorithms.Tasks such as Classification and Prediction can be much more efficiently handled by Deep Learning techniques.Such techniques are essential when we handle data in Big Data scale especially in the context of Fog computing ecosystem.


  • Fog Computing
  • Fog Architectures
  • Fog Applications
  • Fog Security issues and solutions
  • Fog Privacy issues and solutions
  • Fog Applications: Smart Cities, Smart Villages, Smart Homes, Smart Agriculture, Smart Health
  • Fog Application Management
  • Latency aware application design and management
  • Load balancing, Migration and Resource Coordination
  • Stream Data Processing in Fog
  • Fog Analytics and Edge Analytics
  • Machine Learning for Stream Data
  • Processing and Big Data Analytics
  • Deep Learning for Big Data and streams
  • Deep Learning techniques for high dimensional data: Feature reduction
  • Deep Learning for Fast Information Retrieval
  • Deep Learning for Semantic Tagging, Semantic Indexing
  • Large scale models

Please E-mail your proposals to:
C.S.R.Prabhu, Chairman