Smart Sensor Architecture

Smart sensors are sensors with integrated function logic functions, two-way communication, make decisions.

There are very convincing advantages of using silicon technology in the construction of smart sensor. All integrated circuits employ silicon technology. A smart sensor is made with the same technology as integrated circuits. A smart sensor utilizes the transduction properties of one class of materials and electronic properties of silicon (GaAs). A transduction element either includes thin metal films, zinc oxide and polymeric films. Integrating electronics circuits on the sensor chip makes it possible to have single chip solution. Integrated sensors provide significant advantages in terms of overall size and the ability to use small signals from the transduction element [1]. The IC industry will get involved in smart sensor if a very large market can be captured and the production of smart sensor does not require non-standard processing steps.

Will Sensor Networks Suite Sri Lankan Agriculture

When deciding on using wireless systems to Sri Lanka various matters have to be looked into before coming to the final conclusion of it. For example a typical sensor board used for these applications costs around two hundred dollars and the sensors it self has prices ranging form fifty dollars onwards. Even though companies say that this is an affordable solution as for a third world country like us this may not be a very cheap solution. If we were to develop a sensor based system for farmers in our country the costs we incur will be great and sometimes may not be feasible when compared to the available solutions. Furthermore in countries like ours the where labor cost is very low a farmer can afford about ten laborers at the cost of a single sensor. So converting form this manual labor system to a high technological solution like sensor networks in our country may not be easy.

In Virginia vineyard deployments the elevation of the vineyard can be used to plant different types of vines using it in collaboration with temperature data so this situation can be applied directly to our paddy fields. Paddy is cultivated in our country in different elevations at different time throughout the year so by using sensor networks we can measure temperature and other factors that affect quality of the paddy and the feed the data in to appropriate agricultural model to find out which paddy should be cultivated were and in what condition does it give us the maximum yield. This system can be equally applied to other agricultural locations such as vegetable fields.

In Sri Lanka most farmers use fertilizers to get the maximum out of their crop and especially nitrogen based fertilizers. Nitrogen helps a plant to grow but according certain researches carried out more nitrogen can reduce the quality of the crop because of a phenomenon called the nitrogen stress. There are optical sensors created to measure the amount of nitrogen incident on a plant by using the reflective spectrum of the leaves of the plant. But the problem is there are no commercially available wireless optical sensors hence the sensing of nitrogen or other phenomenon using optical sensors should be carried out manually through out the field. But this is a good solution if we can tell our farmers when fertilizers are really required rather than applying fertilizers by consensus. This system can be extended to check the water purity before applying it to the crop and recommendations to clean the water if found not up to standard.

Water quality checks can also be applied to the public water systems as well. When considering the above given deployments the Lofar Agro project is in a much more applicable domain to Sri Lanka than the vineyards. Even the Indian sensor network deployment is more practically applicable to us but the projects itself lacks the maturity. When coming to water-management we can use soil moisture probes on the dry zone fields of Sri Lanka and especially in the semi-arid regions where the farmers cultivate using rain water and where very little irrigation is present. It is very feasible system if we could predict the areas of the field that watering is required. This can in turn be beneficial to a country like us because we can then take the maximum out of our water resource. When considering the architecture of networks to be deployed in Sri Lanka it is much better to have a self organizing network compared to data mule systems and table driven protocols. Even tough this is an expensive solution this is a feasible solution in the long term. Because when sensor systems are deployed in our country there is tendency of sensors getting lost or damaged and expansion of fields with time points more towards an ad-hoc network. If we use a table based system we could easily run into trouble with lost nodes.Our country has a large labor pool to draw from and in agriculture there is constant movement through the fields. But the data mule system is not an option because the cost of deploying the sensor network and then also using people as mules is a much larger cost to incur hence this system is infeasible.

Introduction to Hair Simulation

In the area of computer graphics and virtual reality computer generated virtual humans have a prime importance. They are used in many computer graphics and its related areas in applications such as virtual worlds, digital archiving and gaming. Even though methods exist for creating extremely convincing humans, faces and their related animations, modeling hair in its naturalistic beauty and behavior still remain a challenge.This complexity of hair causes it to be hidden in most applications with hats or by other means or it is just simplified to be polygons which diminish its naturalism, even though hair is an important cue of individual recognition. Complexity of hair simulation arises not from an individual hair itself it comes when these individual hairs are considered as a whole in a human head. So at present when considering modeling, styling, simulating and animating realistic looking hair on virtual humans it remains a slow tedious process.

What is Hair and Hair Simulation?
A typical scalp of hair has about 100,000 to 150,000 hairs and in geometry hair is a thin curved cylinder having varying thickness. Each strand of hair can have any degree of waviness form straight to curly and the colors ranging form black to gray due to pigmentation and shininess. Due to the factors such as geometric intricacies of a hair, light and shadow interaction among hair, small scale thickness of individual hair compared to the rendered image, hair to hair and hair-body collisions hair simulation can be dissected in three major parts known as hair modeling, hair dynamics and hair rendering. Even though these processes can be taken into focus independently in real simulation environment these process are interleaved while processing hairs. Hair shape modeling phase deals with the overall geometry of the hair. It does this by creating thousands of individual hair and then altering the overall geometry of the scalp by using density, hair distribution, individual and overall orientation of hairs and complex hair-hair interactions and hair-body interactions. Once this phase is complete the geometry created by this phase is used by the other two phases to produce the final output. The hair dynamics phase addresses the movements of individual hair stands their collisions with other objects as well as with each other. In the field of virtual humans, hair possesses one of the most challenging rendering problems in computer graphics. The difficulties arise in hair rendering due to the amounts of hair, detail of geometry of an individual strand, complex light and shadow among the hairs and their small thickness when compared to the rendered image. Hair simulation is a highly a research oriented area and is still in its infant state when compared to other areas in computer graphics. There area many research efforts focused on hair simulation. Some of them focus on one aspect such as hair modeling, dynamics or rendering where as others were able to take the general problem into scope. Researchers has divided the current hair simulation models into four categories upon the under laying structure of the above said phases. The four main categories which hair simulation can be divided into are particle systems, volumetric textures, explicit hair models and cluster hair models[1].

So as for an introduction it is fair to say that hair represents one of the most fascinating creations of the Mother Nature which presents one of the most intriguing, mathematically complex and open research areas in computer graphics where an optimal solution even with the latest advancement of graphics based hardware is far from reach.

References
[1] Nadia Magnenat-Thalmann, Sunil Hadap, Prem Kalra, State of the Art in Hair Simulation, MIRALab, CUI, University of Geneva.