ISSNIP

Background

To support the requirements of distributed sensor networks, sensors must possess greater functionality Sample imagethan simply gathering data and blindly transmitting the data to a centralized sensor node. Intelligent sensors are an extension of traditional sensors to those with advanced learning and adaptation capabilities. The system must also be re-configurable and perform the necessary data interpretation, fusion of data from multiple sensors and the validation of local and remotely collected data. Intelligent sensors therefore contain embedded processing functionality that provides the computational resources to perform complex sensing and actuating tasks along with high level applications.

The functions of an intelligent sensor system can be described in terms of compensation, information processing, communications and integration. The combination of these respective elements allow for the development of intelligent sensors that can operate in a multi-modal fashion as well conducting active autonomous sensing.

Compensation is the ability of the system to detect and respond to changes in the network environment through self-diagnostic routines, self-calibration and adaptation. An intelligent sensor must be able to evaluate the validity of collected data, compare it with that obtained by other sensors and confirm the accuracy of any following data variation. This process essentially encompasses the sensor configuration stage.

Information processing encompasses the data related processing that aims to enhance and interpret the collected data and maximize the efficiency of the system, through signal conditioning, data reduction, event detection and decision making. This may involve a collection of filtering and other data manipulation techniques together with advanced learning techniques for feature extraction and classification in order to provide the most relevant data in an efficient representation to the communications interface.

Communications component of intelligent sensor systems incorporates the standardized network protocol which serves to links the distributed sensors in a coherent manner, enabling efficient communications and fault tolerance. Traditional task specific sensor systems often contain a number of limitations in terms of complexity, cost and flexibility. Intelligent sensors aim to overcome these limitations through the utilization of standardized transducer interfaces and communications protocols, resulting in autonomous, distributed, re-configurable sensors.

Integration in intelligent sensors involves the coupling of sensing and computation at the chip level. This can be implemented using micro electro-mechanical systems (MEMS), nano-technology and bio-technology.
A hierarchical structure can be used to describe the functionality of the system, where the lower layer performs the signal processing functions, the middle layer performs the information processing and the upper layer performs the knowledge processing and communications.

Validation of sensors is required to avoid the potential disastrous effects of the propagation of erroneous data. This is different problem than overcoming individual sensor failure. A control system operating decisions made on faulty data can lead to unpredictable behaviour or even complete system failure. The impact of such errors may be reduced through the use of a dense sensor network. The incorporation of data validation into intelligent sensors increases the overall reliability of the system. So an effective means for performing this function is required.

Two approaches are analytical redundancy and hardware redundancy. Analytical redundancy utilizes a mathematical model that compares the static and dynamic relationship between sensor measurements and effectively determines the expected sensor value. The computational expense of this approach can become prohibitive as the number of sensors and model complexity is increased. Hardware redundancy may involve the use of additional sensors and selection of data that appears similarly on the majority of sensors. This approach is not applicable, however, in cases where the presence of an excessive number of sensors has a detrimental effect on the given environment. Knowledge based systems are one alternative, where an intelligent sensor incorporates expert systems that apply reason and infer the solution.

Data fusion techniques are required in order combine information from multiple sensors and sensor types and to ensure that only the most relevant information is transmitted between sensors. Consequently, the load on network bandwidth is kept at an acceptable level. The area of sensor fusion can be approached from a variety of perspectives. Biological science has been used to consider how sensor fusion is accomplished, while cognitive science has explored why sensor fusion is an integral part of perception.

A number of fusion modes can be used to describe the relationships between sensors, These include:

Appropriate sensor techniques should apply the most suitable mode of operation based on available sensor data and given operating environment.

Significance / Benefits

Intelligent sensors operating in a task specific manner with effective data collection techniques enable the development and application of more flexible sensor networks that efficiently utilize and coordinate the limited resources of each individual sensor. By focusing resources according to the state of the surrounding environment and on the immediate task, more efficient operation of the sensor and is ensured.

Accuracy: An intelligent sensor will incorporate features that enable it to compensate for systematic errors, system drift and random errors produced due to system parameters or the characteristics of the sensor.

Reliability: The incorporation of data and sensor validation techniques to detect corrupted data, self-testing of network path connections and sensor operation, as well as calibration of sensor drift, provides yet another level of system reliability in addition to techniques already applied in the network design.

Adaptability: The processing parameters of an intelligent sensor system should be determined automatically and adopted by a higher level in the system architecture. This enables the optimization of the measuring and processing operations, as well as enabling the sensor to adequately respond to changing environmental conditions.

Challenges

Recalibration: The development of a system that accurately determines the type and level of recalibration required by a given sensor.
Information processing: The development of advanced adaptive techniques to improve the efficiency of the data processing and transmission as well as the reliability of the unsupervised learning and decision making elements. The adaptive determination of the necessary and sufficient conditions for feedback of data to other sensors is also required.

Data fusion: Efficient integration of sensors through closed loop control. Given the different contributions made by sensors in a distributed network, appropriate feedback and reconfiguration is required to determine the appropriate combination of data obtained from the respective sensors.

Validation: Knowledge based validation systems that address real time performance and the development of reasoning methods under uncertainty. A real time system must compromise between performance and precision, as well as provide and answer when required even if the generated solution is sub-optimal. Given that the sensor information obtained is considered to have an element of unreliability, uncertainty must be incorporated into the model for data validation.

Integration: The exploitation of emerging technologies such to develop more advanced nano-sensors and bio-sensors that incorporate the desire functionality for application in an intelligent sensor network.

Applications

The following are some existing research applications (refer links below):
Automatic Assembly: Engineering components and assemblies could be made more like organisms in there ability to self-assemble, thus having a significant impact upon production speed, capacity and complexity. Self-repairing abilities are an obvious side-effect of such abilities.

Microendoscopy: The ability to navigate micro or nano-structures through the human body has the potential to make a significant impact on modern medicine. Understanding the mobility and sensory systems of parasites, worms and insects, may provide the necessary design information to realize this objective.

Intelligent suspension for automotive applications: A basic tuned suspension unit requires a spring and damper unit. Tuning the suspension to a particular frequency is generally overlooked. Vertebrate muscle maybe mimicked to construct a high displacement and high damping spring system.

Gels: The principle of a fluid enclosed in a membrane being made to do useful work can be seen in our own muscular system, plants and in the skin of worms. Contracting the muscles in the body wall and increasing its internal pressure the worm is able to change shape. Controlling the swelling and contracting of a polymer gel appropriately encased, enables a system to work as an artificial muscle.

Smart fabrics: Analyzing the insulation layers of animals and other natural responses to temperature fluctuations may contribute to the development responsive clothing, with properties based on the state of activity of the wearer. This would reduce the number layers required by the wearer while remaining suitable for a variety of weather conditions.

References

  1. Ibarguengoytia PH, Sucar LE, Vadera S. Real time intelligent sensor validation. [Journal Paper] IEEE Transactions on Power Systems, vol.16, no.4, Nov. 2001, pp.770-5. Publisher: IEEE , USA .
  2. Andrew J. Fry. Integrity-Based Self-Validation Test Scheduling. IEEE Transactions on Reliability , vol. 52, no. 2, June 2003, pp.162-167.
  3. Robin R. Murphy. Biological and Cognitive Foundations of Intelligent Sensor Fusion. IEEE Trans. On Systems, Man and Cybernetics   Part A: Systems and Humans , vol.26, no.1, Jan 1996, pp. 42-51.
  4. J. C. Patra, A. C. Kot and G. Panda. An Intelligent Pressure Sensor Using Neural Networks. IEEE Trans on Instrumentation and Measurement , vol.49, no.4, Aug 2000, pp 829-835.
  5. A. Sachenko, V. Kochan and V. Turchenko. Instrumentation for Data Gathering. IEEE Instrumentation & Measurement Magazine , September 2003, pp. 34-41.
  6. Ricardo Gutierrez-Osuna, Intelligent Sensor Systems, Wright State University
next
top of page