N-SMARTS: Platforms

Home
Platform
Data
Advanced Sensing
Participatory Urbanism
Publications
People
Internal Wiki
Source Code
Links
Because we don't have a clear picture of what the pollution data and movement patters of users will be, we need to gather data up front, before we have an integrated sensor/phone platform available. For that reason, we have put together a portable sensor platform which can be carried around, allowing a person to gather data that is roughly similar to the data which will be gathered by the integrated platform. The data acquisition platform will allow us to develop and test the algorithms that make up the core of the N-SMART platform.

The Data Acquisition Platform

The personal sensor platform

Assembling the automotive sensor platform
The data acquisition platform consists of off-the-shelf pollution sensors, and a GPS. Each unit contains
  • A Lascar EL-USB-CO Carbon Monoxide data logger
  • A Garmin Qwest GPS (with external antenna)
  • A NO2, SO2 or O3 datalogger from BW Technologies
All of the devices log data, and their clocks are synchronized so that the data from each device can be correllated. See our wiki page for details on the sensors.

Personal Sensor Platform

The personal sensor platform is a clip and bar on which the GPS and the Lascar sensor is mounted, and on which one or more of the BW Tech sensors can clip. The clip is used to afix the platform to the user's belt, pants, etc. We use the personal sensor platform to most closely approxmiate the environment to which the integrated platform will be exposed.

We currently have 4 personal sensor platforms in deployment in Accra, Ghana, West Africa. The sensors are being carried by undergraduate students at Ashesi University, located near downtown Accra.

Automotive Sensor Platform

Since cars move quickly around a city, an automotive platform, particularly when attached to at taxi or public transportation vehicle, has the potential to gather a significant amount of data about the pollution in a city, although it will be biased towards pollution near roadways.

The automotive platform, however, faces some challenges which a personal sensing platform does not. First, the platform is not necessarily immediately under the physical control of the driver, and in some place (including Accra, Ghana), thieves often reach in to slow-moving cars and snatch valuable looking electronic devices! Thus, the sensors should be hidden inside of an opaque container.

Secondly, the container needs to allow airflow through it, so we have used a tube with vents on either end to allow air to flow freely. Third, the pollution levels inside the car might not be representative of the pollution outdoors, so we mount the tubes facing slighly out the window, attached to the handles above the door. This also encourages airflow through the tube.

We currently have 6 automotive sensor platforms in deployment in Accra, Ghana, West Africa. The sensors are being carried in taxis from the local union.

The Integrated Platform

An early prototype integrated sensor board

The integrated battery pack, with the original battery in red, the PCB in green and the new cover translucent.

The integrated battery back, with the cover shown as opaque.

The integrated platform is designed to approximate a phone manufactured with sensors integrated directly into the phone itself. This model will allow significant cost reduction with respect to less complete integrations. Rather than actually manufacturing a new phone and enclosure, however, we simply replace the battery pack of the phone with a module that clips in to the battery well of the phone, and contains both a battery and the sensor module. The design pictured at the right was created by Kyle Yeates, and will be manufactured using We have two main requirements for the phone we use for the integration. First, it must support fast, reasonably accurate location fixes and should support indoor location fixes as well. Secondly, it must have a programmatic interface for communicating with the sensor module. For our first integration attempt, we have chosen (the unfortunately expensive) LG VX9800.

Location fixes using AGPS

The AGPS system integrated into phones with Qualcomm chipsets has a significant advantage over other AGPS systems because of the synchronized clock of CDMA devices. In combination with the standard ephemeris information provided by the network, Qualcomms AGPS systems gets the equivalent of 20-25dB of gain on the GPS signal over a standard GPS. Under many real world circumstances, this translate in to cold fixes in less than 10 seconds, hot fixes in less than 2 seconds, and fixes indoors. The integrated localization solution also allows for coarse granularity fixes based on triangulation from the base stations when a GPS signal is not available.

Fast fixes and indoor fixes give the device a degree of autonomy not possible with other localization techniques. Fast fixes enable the device to sample without user intervention. Otherwise, the user must seek out locations with a clear view of the sky in which to take a sample, and often must remain in that location for a matter of min-utes to get a location reading. This usage model would clearly limit participation to only the most dedicated and active of users. Qualcomm's AGPS solution, on the other hand, enables passive participation, requiring no change of behavior from the user, and thus enabling the system to potentially gather information on behalf of mil-lions of users.

Communicating with the sensor module

Several new BREW phones now enable the BlueTooth Object Exchange (OBX) protocol, which gives us a cleaner interface to the phone than a phone-specific RS232 port. Power management with a bluetooth radio will be an important issue, and we are investigating how to duty cycle the radio in order to conserve power and limit reading latency.

Our module also supports USB, if it is required, although the latest version of the board does not include a header or line leveler for USB.

The sensor module

The v1 sensor module contains: The board also has a MSP430 microcontroller and other peripheral circuitry. The temperature sensor is to calibrate the readings from the CO sensor, since the CO sensor's gain is slightly affected by differences in temperature. The accelerometer will assist with inferring the activity of the phone user, which will in turn provide more accurate models, besides being interesting in it own right.

The first board revision was larger than necessary (Approximately 1.3" x 2.5") in anticipation of design and layout errors, which are easier to fix with more free space. The next board will be significantly smaller, at around 1.2" x 1.6", and could be made even smaller with smaller surface mount components, using the back side of the board and even tighter spacing.

The enclosure

In order to get a tight integration with the phone we will replace the battery module with an enclosure which fits into the battery well of the phone, and which contains both the sensor module and a battery. The enclosure will have an orifice which will expose the CO and temperature sensors the atmosphere, and an RS232 connector which will attach to the serial port of the phone. The initial enclosure will be manufactured using the rapid prototyping and solid freeform fabrication facilities at the Ford Lab, run by the Berkeley Manufacturing Institute. This these tools allow arbitrary 3-dimensional forms to be printed using technology similar to an ink-jet printer. For larger runs, the BMI also has injection molding capabilities. The design pictured at the right was done by Kyle Yeates.