Foster Information Ubiquity

Debasis Bandyopadhyay, Head, TCS Innovation Labs - Kolkata

Ubiquitous computing seamlessly integrates digital and physical devices by making many computers available throughout the physical environment, while making them effectively invisible to the user. Users can access digital data and applications from the environment as easily as, (if not easier than) accessing them through their personal computers. Since ubiquitous computing exists in the user’s environment, the technology is sustainable if it is invisible to the user and does not intrude the user’s consciousness. This requires functioning of a multitude of devices in the environment to be oblivious to the users. Mark Weiser in his paper, “The Computer for the Twenty-First Century” (Scientific American Magazine, pp. 94-100, September 1991,) defines ubiquitous computing as a technology that weaves itself into the fabric of everyday life until it is indistinguishable from it.

The areas of focus at our Innovation Labs extend across the following areas:

• Focus on four different types of Platforms
  • Sensor Nodes:
  • Aggregator platforms
  • Mobile / Portable visualization platforms
  • Large storage and distributed computing

  We also carried out focused research in the following problems areas that interest us:

• Interoperability: With the dramatic increase of connectivity requirements and diverse range of devices to be dealt with, there exist several gaps in the standardization effort towards scalability and interoperability in ubiquitous computing. Our focus in this area is more in the upper layers of communication as compared to physical and data link layers.
• Context Awareness and Adaptability: Context Awareness is the ability for applications to sense and understand end-user information We are working on designing a domain-agnostic analytical framework through advanced statistical processing. Context-awareness and adaptability, together, form the equivalent of a closed loop feedback control system. We are working on converting the feedback logic into “actuation” of user environment with minimal user intervention.
• Privacy and Security: The requirement of context awareness and adaptability by virtue calls for sharing of end-user data across multiple devices which, is in direct conflict of the end user's privacy and security. With so much personal information becoming available over computing devices and the networks connecting them; ensuring privacy of user data in a ubiquitous environment is a significant challenge.
• Robustness: Failures, errors and context changes need to be handled efficiently in order to guarantee the system performance within expectation limits. Schemes for error handling, fault tolerance, co-operation and intelligent decision making are to be incorporated into systems to address these challenges. Robustness is, therefore, an important attribute in ubiquitous computing as it contributes towards ensuring a richer user experience

In the current year extensive work has been done on the following areas:

• Home Infotainment Platform as a Ubiquitous Home Gateway The Home Infotainment Platform (HIP) is a generic multimedia framework based platform that can decode and render multimedia and information content from the internet.
• Connected TV for Content Analysis. Connected TV uses sophisticated multimedia processing techniques to understand the context of the broadcast TV video and audio. Once the context is extracted, several valueadded applications can be provided to the end-user.
• Complex Event Processing Context Awareness enables automatic detection of the user and usage scenario based on real-time processing of sensor and network event streams and thereby enables a rich user interaction and experience. Complex Event Processing is one technique of discovering context and is therefore an important capability that we are building into the TCS ubiquity story.
• Open Storage Solution While streams processing is used to perform analytics on data that is “in-stream”, data still needs to be stored in a storage media for traditional data mining and analysis. This area has received fair amount of interest both academia and industry. In this regard the following are of importance:
  • Creating low cost, high availability storage infrastructure for sensor data using commodity hardware
  • Using clustered file systems for building massively scalable file serving infrastructure
  • Building parallel data warehouse infrastructure for large scale data mining tasks
  • Using programming models such as Map-Reduce along with clustered file systems for performing parallel number crunching tasks
  • Error Control and Erasure Coding In order to address the requirement for robustness requirement, advanced error control & erasure coding schemes for adaptive error correction and error tolerance in the communication and storage is of substantial value. A multi-application framework for multisensory data fusion is under development.