Project Proposal for Tiffany

 

SUPERB Summer 2001

 

Background:

 

The PicoRadio project is a research effort exploring dense networks of very small low-power System-On-Chip (SOC) devices. These devices communicate with each other via radio links.  The application space for PicoRadio includes numerous types of sensor networks and anything else that requires large numbers of inexpensive devices producing relatively low-bandwidth data in a decentralized way.

 

In order to build a system of this nature, we have to define very clearly how the network will work i.e. what the pieces are, and how they fit together.  A common way to do this is to construct a high-level model of the system.  High-level means that it describes network function completely but in a very general way - no unneeded details such as what kind of radios to use or the specific way data is transmitted between nodes.  A model is a formal description of a system that contains enough information to completely describe how the system works.  In our case we are viewing the system at a high level, so the model tends to be symbolic, that is, a device (or node) may be represented by a rectangle in a diagram and its behaviour may be described using 'C' code or a finite state machine.  Communication between nodes may  be represented by lines between rectangles and communication behaviour may also be described by finite state machines or some other of "model of computation".

 

We use a software program called VCC (Virtual Component Compiler) to create our models.  VCC allows you to draw the symbolic pictures an the relationships between nodes.  It also allows  you to "execute" the model i.e. provide a stimulus (sometimes called an event) at some input and see how the system described by the model would behave under the same circumstances.

 

We already have a pretty good idea of how a PicoRadio system would look (its "architecture").  We're currently in the process of building an executable VCC model of a likely sensor network.  This network has three primary node types, each with its own role.  These are: the sensor, the controller, and the actuator.  Simply put, the controller asks for data from a sensor (maybe a temperature sensor, or humidity, or whatever).  The sensor responds with a sample.  The controller has to know which sensor to ask, where that sensor is, and how to ask it in order to get the right data.  The action of requesting a sample may seem simple on the outside, but there are complexities in this dialog that the model helps us to understand.

 

Actuators are similar to sensors except that instead of sampling a physical attribute (read-only), actuators "act" on the physical world (write-only).  For instance, an actuator may be a mechanical solenoid that moves a robot arm, a remote light switch, or an air conditioner control.

 

Currently, we have a team of two undergraduate researchers under the guidance of graduate students working on the model.  They are focusing on the controller and sensor nodes and the communication between them.  The actuator node also needs to be modeled.

 

 

Project Outline:

 

This project will be the VCC modeling and simjlation of a PicoRadio actuator node.  

 

It will follow these steps, more-or-less:

 

  1. Understand the PicoRadio network architecture.
  2. Learn the basics of  VCC.
  3. Work with Ruth and Gerald to define an actuator/controller interface and plan division of labor.
  4. Develop a model for an actuator that interfaces with the controller.
  5. Simulate a PicoRadio system with multiple actuators, one or more controller, and possibly sensors.
  6. If time permits, implement the actuator on the PicoRadio Test Bed.

 

We’ll work together to refine these steps along the way.

 

What you will learn:

 

  1. A widely used methodology for complex system description.
  2. An architecture for advanced, ubiquitous, low-power networking.
  3. A powerful cutting-edge modeling and analysis tool..