Is it the phone or the service provider?

Thursday, March 15, 2007

Is it the phone or the service provider? The last post seemed to suggest that it is not always bad service that results in bad user experience. This note attempts a justification for that statement.
Let's first look at it from a service provider's perspective. The economic argument is fairly simple. They have the spectrum and they have a certain Capital Expenditure budget. So, they design the network (i.e. placing cells, and adding carriers and channel elements) based on peak user traffic and capacity requirement. Initial cell-sites may be set up such that each site supports upto a certain number of users and their placement is such that areas with dense activity get more cell-sites. Now, when this maximum user limit is hit, then channel elements and carriers are added to each of these cell-sites. Sporadically, the peak capacity is achieved due to a burst of activity and is likely to be ignored until the frequency of occurence of such an event in that given site is beyond a pre-set value. This is when you may experience a dropped call or when your call does not get through for a few minutes. And of course, there may be spots where the coverage is not good enough to get a call through owing to environmental conditions, buildings/tree canopy etc. Under this condition, your mobile is eternally searching for a signal or your call is lost with the message "Signal Faded".

This being said, the question now is if the phone manufacturer can go wrong at all. There is often a misconception that the standard bodies like 3GPP or 3GPP2 define the transmitter and the receiver in their entirety. This is not true. Typically, only the transmitter structure is defined clearly. Infact, that is where most of the intellectual property (IP) wars are fought in the standards. The receiver design is implementation specific and each company will have its own sweet technique, trade secrets and secret sauces to get performance edge. As far as receiver-side standardization is concerned, only minimum performance requirements on a limited set of test cases is stipulated. How a vendor meets it is their issue. For example, in 3GPP, there are type 0, 1, 2 and 3 receiver requirements which were arrived at using a simulation-calibrated rake receiver, receive diversity, equalizer, and receive-diversity+equalizer respectively. Again, the minimum requirements are arrived at after aligning simulations with the so-called "implementation margins" and an average of the submissions is taken as the minimum specification.

Now, the standard however does not specify that a type 2 receiver should be an equalizer. It also does not specify the maximum performance or average performance expected. Someone can obtain a type 2 performance using a 2-antenna solution or even a rake with a better decoder. So, it is here that companies try to go one up on the competition. On the flipside, vendors are also likely to minimize their costs by barely meeting the specifications. After all, an increased performance in the voice case does not impact the end user. Advanced receivers in this case, serve to reduce the transmit power needed to achieve the same link budget thereby increasing the cell-capacity. This can in turn translate into Capacity Expenditure savings for the service provider, especially if the infrastructure is not completely in place or if the network is not over-designed. But service providers again have to strike a balance between capacity and user experience. They are likely to attract more customers and thereby more revenue (though at lower margins) if they can have an array of sleek and multi-functional phones than if they have a really clunky-looking but great performing phone. How would the user care? For example, most of us would find it ugly to have two-antennas sticking out of our cell-phones. So, vendors try to embed the second antenna into the body. And that not being optimal in terms of its correlation properties, does not give expected performance.

Phones that barely meet the specifications can have issues that are not apparent in lab tests but are specific to the field. For example, 3G phones may have problems with backward compatibility with 2G networks. So you may face problems when you are moving from a WCDMA infrastreucture to a GSM infrastructure. Similarly, different phones are likely to handle situations like hand-offs (the process of shifting the mobile from one cell to another). So you may lose a call due to an inferior procedure that barely meets the requirements. There are many issues that are less apparant as phone problems. As mentioned in the previous post, the service provider is very likely to take the blame for much of this.

So the next time you have a service issue, think twice before you blame your Sprint or Verizon connection. It could just be because of your attraction to fancy devices which also serve as mobile phones.

Posted by Vijay Nagarajan at 11:02 PM  


Post a Comment