At Skyhook Mr. Rice is responsible for identifying, creating and building new market opportunities for its Wi-Fi positioning technology through direct and indirect sales channels in the US and overseas. Previously, he created, launched and led the Self-Service OnDemand business unit at Siebel Systems. Mr. Rice joined Siebel as part of the acquisition of edocs where he had held a range of key positions managing new market and business development initiatives. Prior to edocs, Mr. Rice was a corporate attorney; he also served four years as an officer in the US Army. He holds a BA in International Affairs from Georgetown University and a joint JD/MBA from Boston College.
Jed Rice: Our Wi-Fi Positioning System is a software-only location platform that provides up to 20 meter positioning accuracy to any Wi-Fi enabled mobile device.
It basically works as follow: Wi-Fi base stations repeatedly broadcast a signal beacon announcing their existence to the surrounding area and these beacons typically travel 150-200 meters in all directions. Our software identifies all of these existing Wi-Fi signals and our algorithms calculate your location.
To enable this solution we have created a reference database of known Wi-Fi access points. Like a Tele Atlas or a Navteq, we have a fleet of about 200 data collection drivers who drive every street, highway, road, etc… we also use bikes and people who walk to collect this data on places where you cannot go with a car. This approach gives us a reliable and uniform infrastructure of reference points that allows us to deliver consistently accurate positioning.
GPS BN: OK, but Wi-Fi access points are changing over time, so how do you update this database?
JR: There are a couple points I want to highlight about that. First it is very critical that we do the baseline scanning of the Wi-Fi points because we are offering a commercial solution and our clients want a reliable system. Baseline scanning means that we offer performance levels agreements – in terms of accuracy and availability – that other providers can't offer. So with that baseline scanning we can ensure universal coverage and a consistent performance.
Second, once we have done this reference database we can update it automatically. Every time a device or application calculates location using Skyhook’s technology, for example, every time an AOL Internet Messenger user looks for buddies nearby, he is sending us information about how the Wi-Fi landscape looks around him. And this information is fed back into our system.
JR: Obviously the more hotspots you have the better the accuracy of the position. But we provide pretty extensive coverage, for example in the United States in 2006 our drivers drove 1.5 million miles in nine and half months. We now cover 70% of the US population and 2,500 cities – to include small towns like Middlebury, Vermont. That's over 17 million access points which we have collected and verified ourselves. And because we do a thorough and comprehensive data scan, we have between 150 and 500 readings of each signal provided by an access point. Because of that we can do 40 meters accuracy 95% of the time with just three access points. In Western Europe we will have something similar in the next 5 months and within 9 to 12 months in Asia.
GPS BN: I have got a last question about your technology, is it network dependent? Do you need a connection to make it work?
JR: This is not compulsory. Our technology works in three models. One is autonomous which means our software is running on a mobile device and the reference database is also on the device. This is the case for the W10 Personal Mobile Player (PMP) from iRiver. We can put the 17 million access points we have in the US and compress it down to under 120 Mb. So for an iRiver device with 1 GB storage this is nothing. And we can provide extended coverage over time as well as data updates when users connect their device on a periodic basis. In the case of the iRiver product, iRiver will have a link with its customers through their web where people can get songs, videos, etc. and get an update of our data at the same time.
The second model we offer is pure network-based. So on the device itself is a very thin software client, under 75 Kb, and every location transaction does go over the network. In this case the production server, sitting within either a carrier operation network or our own, performs all the calculations and sends back the position.
The third model which is also designed for mobile phones is a dual mode. In this case I have a subset of the reference database on my device. I am based in Boston so I have the entire New England access point reference database on my phone: this is perhaps a few hundred Kilobytes. When I get off a plane in San Francisco, the software does not have the database for this part of the country so it is going to do a network-based location transaction, but depending on the settings on the phone, it might also download all the data for San Francisco so that the next location could be done autonomously.