I thought I'd present some casual observations I made throughout the day Wednesday on a trip from Portland to Seattle, as well as some newly reported information about the AT&T 3G network that's hit the 'net over the past 24 hours or so.

The back-story here is that I - like many others - have found the reliability and consistency of the iPhone 3G to be less than satisfactory while on the 3G AT&T network.

First of all, it became clear to me over the course of several hours yesterday that the iPhone is not to blame with regards to connectivity on the 3G network. While driving from Portland, Oregon to Seattle, Washington and back yesterday, I had the opportunity to run a whole slew of speed/connectivity test sessions using the iPhone app called "iNetwork Test" (click here to get the free app in the iTunes App Store).

AT&T actually has fairly impressive 3G network coverage from south of Olympia, Washington practically all the way to Seattle, with one or two small gaps in-between where the phone switched to EDGE. Much of the area along that I-5 corridor is rural or sparsely-populated. From a wireless connectivity standpoint, it's a pretty decent area to live in if you're going to be far away from the city.

My experience in using the 3G network along my drive up and down the Interstate can be summed up thusly:

In areas with higher population density, and thus more iPhone (and other device) users, ability to a) connect to the voice network and make calls, b) stay connected to the voice network, c) make data connections and d) maintain data connections was substantially worse. The difference between dense and sparsely populated areas was like night and day.

Where population density was lower, even in cases when fewer bars are displayed on the signal strength icon, voice and data connections were reliable and solid without exception. In contrast, in high-population areas even full-signal connectivity was spotty and unreliable.

I'm running the latest iPhone software, v2.0.2, which both Apple and AT&T have encouraged people to upgrade to. AT&T even sent a text message to all users asking them to upgrade - a first-time action on the part of the carrier.

Some new information, part of which you'll find quoted below, helps explain why I experienced substantially poorer performance in the cities and heavily-populated areas but not in the rural sections of my drive. According to reports, it appears AT&T's 3G radio systems are power-constrained, and are not able to maintain all the connections. The incredible number of iPhone 3G devices on the network - especially in metropolitan and urban areas - is most certainly placing a heavy load on the radios. In addition, iPhone 3G devices that have not been updated to the v2.0.2 software are placing an even heavier burden on the radios from a power-consumption standpoint.

So, there's a power-management problem, as well as a capacity problem. When the network "noise" in the radio spectrum used gets to be higher, the towers have to increase power to try to overcome the noise. You can see how that doesn't work. Eventually the noise keeps climbing and the power consumption at the tower (and presumably on the iPhone as well) goes through the roof.

More towers would increase capacity, reduce power requirements and resulting noise, and generally improve coverage. But that's not something that can be changed overnight.

All of this helps explain why my ability to make calls, connect to the 3G data network and download at high speeds was much better where the network is only lightly used.

The Daily Tech site has a detailed report (and some intelligent reader comments) that describes the cell-site power issues, the problems related to the older iPhone 3G software, and other items. Go to the Daily Tech site to get all the details. Here is a portion of the information, including some text quoted from Roughly Drafted Magazine, whose author was able to get some new details from a source inside AT&T's wireless business describing the power issues and what the iPhone's v2.0.2 software update changes:

Basically the update "fixed power control on the mobile" according to the source. To understand what they're going to say next, you must first know a bit about AT&T's jargon for UMTS -- the technology it uses to deliver its 3G network. In the technology, phones are referred to as user equipment, "UE" for short. The base transceiver station towers are known as "Node B".

With this jargon in mind, the AT&T source explains:
"In UMTS power control is key to the mobile and network success. If the UE requires too much downlink power then the base station or Node B can run out of transmitter power and this is what was happening. As you get more UEs on the cell, the noise floor rises and the cell has to compensate by ramping up its power to the UEs. If the UE power control algorithm is faulty then they will demand more power from the cell than is necessary and with multiple users this can cause the cell transmitter to run out of power. The net result is that some UEs will drop their call. I have seen the dropped call graphs that correspond to the iPhone launch and when the 2.0.2 firmware was released. The increase in dropped calls, (were the result of) dropped calls due to a lack of downlink power."

In essence, the iPhone is asking for a stronger signal than it needs. In areas with lots of users, some or all of whose phones are doing this, calls start to get dropped and signal quality drops. This all follows with the conclusions the media had reached -- the problems were somehow correlated to user distribution and seemed puzzlingly to be both with AT&T's network, and with the hardware.

The source continues:

"The power control issue will also have an effect on the data throughput, because the higher the data rate the more power the Node B transmitter requires to transmit. If the UEs have poor power control and are taking more power than is necessary then it will sap the network’s ability to deliver high speed data. This is one of the reasons why AT&T has been sending text messages to users to persuade them to upgrade to the 2.0.2 software. In a mixed environment where users are running 2.0, 2.0.1, and 2.0.2, the power control problems of 2.0 and 2.0.1 will affect the 2.0.2 users. It is not the network that is fault but the interaction of the bad power control algorithm in 2.0 and 2.0.1 software and the network that is at fault. The sooner everybody is running 2.0.2 software the better things will be. Having seen the graphs the 2.0.2 software has already started to make difference."

Since transmitting lots of data takes lots of transmission power, and transmission power was unnecessarily being raised above that necessary for the use levels on phones, the network in areas of heavy use was unable to handle high speed data.