Understanding PDE (Positional Determining Entity)

In the past 2 years, the industry has finally moved beyond LBS as a category to LBS as a feature. As stated previously, “What app wouldn’t be more useful with location?”! What’s misunderstood in all of this is how location is retrieved by the app and what this means.

E911 (government mandated subscriber location lookup) in the early days was the initial motivation for opeartors to introduce network-based location lookup. This location is acquired through a combination of network triangulation technologies and is delivered to the requested entity (eg person or app) via a positioning server. This approach is often referred to PDE or carrier-based location lookup.

The challenge with PDE has been in business models; operators have been charging for PDE location queries and a variety of middle-men and/or aggregators have been trying to sell this access (eg Wavemarket, Autodesk’s former mobility group, Alcatel Lucent and others). Most developers obviously can’t afford to pay 5-10 cents per location lookup and have thus resorted to the many free alterantive ways to get location such as determining location by the cell tower or getting location through WiFi or most obvious, getting location from GPS (readily available through APIs in smartphones today.

You may ask, why PDE if you can get location in other means? The challenge with the alternative solutions, is that the app can only get location for you (and you only). This is great for use cases like Google Maps but what if you wanted to get the location of your buddy or what if you wanted to send an SMS when one of your friends walked by your house – yes, that would be kind of creepy but advertisers love this scenario (proximity marketing). Unfortunately, the only way besides doing a FourSquare check-in to get the location of one of your buddies or a passerby is using carrier-based location lookup.

To entice more developers, many of the carrier location aggregators have been trying to offset the 5-10 cents per query pricing via alternative business models. For example, a developer can receive subsidized queries in exchange for sharing advertising revenue from their app. Unfortunately as much as mobile advertising is growing, it still cannot offset even a 5 cent query. The better question to ask is why is there a charge at all – if PDE was free, there would be a whole new set of apps that could be enabled built leveraging real-time location. For example, without needing an app running in the background, I could track the entire life-path of a user – an advertiser’s goldmine in terms or profiling! Well, as with most things, the reasons things are the way they are is often not the most logical:

1. Without naming operators, some can’t scale PDE. I’m pretty sure they want to open it up, assuming the right privacy features were in place but the query volume would kill the network infrastructure. Let’s see what happens as initiatives like GSMA’s OneAPI continues to push forward – this is a solvable problem.

2. Some operators are restricted by their licensors. For example, CDMA operators use Qualcomm’s QPoint to power the PDE and likely have licensing fees based on volume that inhibit their capability to open it up.

3. And of course the biggest problem is that there are a handful of developers (eg Family Locators) who are actually paying 5-10 cents per query. Unfortunately, when somebody is willing to pay, then let them pay even if it inhibits innovation. I guess we’ll have to wait for Google to make it free (somehow)!

In any case, let’s see how this evolves. Carrier-based location lookup can open-up a whole slew of new applications that cannot be achieved effectively today. It moves the problem from getting location to the larger problem of privacy which many others are trying to tackle.

Measuring AT&Ts Network

I’ve been meaning to this post for a while – it’s inspried by a similar analysis of Vodafone coverage in Germany via WirelessMoves. Note, I’m definitely not an expert when it comes to carrier infrastructure especially since I’ve spent my whole career on the software side of the mobile biz :)

Anyways, I recently read that it’s estimated that ATT has approximately 50K cell sites in the US. I’m not sure what’s the split between GSM and UMTS base-stations but from a subscriber standpoint, I have read that approximately 30% of US subscribers are 3G subscribers meaning they have a 3G handset and are connected over 3G. In any case, let’s do some simple math assuming all 80M ATT subscribers were connected over the 50K cell sites. This means each cell site is serving 80M / 50K = 1600 customers. Base stations are usually split into three sectors meaning each sector covers 1600 / 3 = 533 customers. Now, the average ATT subscriber uses 760 voice minutes each month so now multiply 766 min * 533 customers and you have 408,278 voice minutes per sector. Now obviously, most of the calls are within a certain set of hours each day, I will assume 16 so you then have 408,278 minutes / 30 days / 16 hours = 850 minutes per sector per hour. A sector is typically equipped with 2-3 transcievers which can each server between 6 and 8 voice calls meaning the potential maximum would be 60 minutes * 3 transceivers * 8 voice calls = 1440 minutes per sector per hour – I have no idea if that is reasonable or not :)

Lots of assumptions here and I’d love to get some more accurate data. It’d be interesting to look at data consumption per sector and see if we are really under-capacity. Obviously, it’s not as simple as 50K cell sites since there is quite a bit of tower leasing meaning only partial capacity as well as a significant number of subscribers being prepaid and/or part of an ATT MVNO (ie Tracfone estimated at 14M prepaid subscribers). We also assuming an even distribution across the country which definitely doesn’t hold, you’d expect a higher density within urban centers.

In any case, interesting as it is – let me know if you have better data.