"Revolutionary networking concepts and an unprecedented mix of technical challenges have made Wireless Sensor Networks (WSN) one of the major research trends of the 21st century. However...despite years of research and development and technical maturity, WSN products and solutions are yet neither fully adopted nor widely deployed."
On April 30, 2008, Thomas Kunz, Director of the Technology Innovation Management (TIM) program at Carleton University delivered a presentation entitled "Wireless Sensor Networks: What and Why?".
The TIM Lecture Series provides a forum to promote the transfer of knowledge from university research to technology company executives and entrepreneurs as well as research and development (R&D) personnel. This conference report presents the key messages and insights from the three sections discussed during Professor Kunz's presentation.
Introduction to Wireless Sensor Networks
The first section served as an introduction to wireless sensor networks (WSN). On the hardware side, sensors range in price and functionality from cheap and unreliable to expensive and mission critical. There are still many engineering challenges yet to be resolved. These challenges include: i) providing sustainable power and overcoming distance limitations in sensors; ii) creating global standards for radiation and privacy; iii) providing sufficient address space for nodes; iv) determining the environmental implications of discarded sensors; and v) studying long term health effects. Designing efficient networks is also challenging as the design must provide redundancy for nodes that fail and sensors require strategic placement in order to provide effective data collection.
Applications for WSN are many and varied and the application possibilities seem limitless. Particularly attractive application areas are smart homes and real-time traffic information. WSN is an emerging technology, providing many business opportunities to engineers who can solve the technical challenges and entrepreneurs who can capitalize on the new markets.
WSN technology raises many interesting dilemmas. Sensors now have the ability to collaborate as peers instead of merely uploading their data to a central server; an example of this is ZebraNet. In WSN where humans are being monitored, such as in medical scenarios and smart homes, privacy of the data collected is critical. Also, will the fact that people know that they are being monitored result in a change in their behaviour?
Distributed sensors that are not under one entity's control are an important trend which raises the question of "who owns the data?" Actuators, mechanisms which introduce motion or which clamp an object so as to prevent motion, have alarming privacy implications.
WSN Applications and Challenges
The second section of the lecture concentrated on the challenges in designing WSN applications. Key insights from this section include:
- when designing hardware, processing is cheap but transmission of the data is expensive
- programming sensor networks is different than programming for the Internet
- database queries for WSN data also differ in their logic, though SQL does provide a well-known abstraction to developers of applications
- sensor data eventually ends up in a data warehouse for scientists to use
- application designers should be aware that currently the monetary value is in the hardware, not the software
- software design is based on current hardware constraints and we may be overcompensating for these constraints; conversely, conservative software design allows us to make hardware smaller
It was noted that some of the technologies developed for WSN will find their way into the Internet. An example provided by the audience was cloud computing as propagated by Google, which seems very similar to the data aggregation techniques discussed in the WSN community.
WSN Networking and Local Research
The final section concentrated on WSN networking. While most networks have been standardized for many years, WSN still offers many design opportunities. Network protocol stack designs are difficult but provide research opportunities. The traditional layered protocol architecture has proven successful in the Internet, but has overhead and redundancy challenges, making it less appealing for constrained wireless/embedded devices. Moreover, much more research is needed for transducers and actuators. Timing (synchronization) and localization are also important engineering challenges.
In addition to network design challenges, realistic simulators are needed for testing sensors, applications, and network protocols. A technology road map for inflection points would be also be useful. The core hardware platforms are becoming cheap, but academic research which pushes the range of applications and develops new sensors is expensive. Distributed wireless networks are still not yet built from off the shelf components for creating one's own research testbed. New hardware such as radios, sensors, and micro controllers are still being developed in research labs.
Two domains have emerged: i) geographically static sensors and ii) moving sensors. The second domain is more relevant for commercial opportunities. The Ottawa area has a range of activities in the WSN domain: i) companies that build sensor platforms; ii) research into sensors and actuators, particularly in the biomedical domain; and iii) work in academia and government research labs on network protocols and building testbeds. What may be missing is research on operating systems for embedded systems and middleware.
After the lecture, the audience held a brainstorming session on possible business opportunities around WSN technologies. A WSN opportunities page has been added to the Talent First Network wiki. Readers are encouraged to add new opportunities or contribute to an existing opportunity. Contacts are provided should you wish to discuss value propositions with the champion of an opportunity.
An FDL'ed Textbook on Sensor Networks