You could recently read that many operators firmly believe that all capacity/cost problems in wireless data access can be solved by replacing their current equipment with new models with higher peak-rates (read LTE). Since higher and higher peakrates in the base stations is matter of more efficient electronics and signal processing, the costs for more bandwidth, will keep approaching zeros and the exponential growth in traffic demand is not a problem. Case closed – problem solved…. or is it ?
Unfortunately not. Look at the graph, that shows the access point capacity in a cellular data system which is the average data rate a user will experience as function of the peak- rate in the system. This is by the way a hypothetical system- operating at the theoretical maximum data rate (at the Shannon bound for comm. theory geeks). There are a number of curves for various cell edge rates – i.e. the worst data rate in the cell. The behavior is very clear, the cell average rate saturates as we increase the peak rates in the cell. This is due to the fact that in a large cell, very few users will enjoy high rates, close to the peak rate. The average rate is by large determined by the edge rate and increasing the peak rates gives only marginal improvement – even INFINITE peak rates doesn’t do you any good. Running a 100Mbit/s LTE system in our numerical example, makes thus little sense if your edge rate is 1 Mbit/s. You will end up with a 5-6 Mbit/s average rate (for that you need basically only the 10-15 Mbit/s peak rate that an HSPA system would deliver) . There are certainly a number of parameters that can be varied (i.e don’t take the absolute numerical example above too serious) , but the general behavior is definitely the same. (A paper with all details is in the publishing process, for those interested in details).
So, sorry, in order to significantly increase the capacity, there are no real shortcuts. The edge rates must be increased in order for high peak rates to become meaningful. Unfortunately, there are no real other options than to increase the number of access points. “Zander’s formula” (stating that cost grows basically linearly with capacity) from 1997 remains valid.
 Jens Zander, “On the Cost Structure of Future Wideband Wireless Access”, IEEE VTC ’97, Phoenix, AZ, May 5-7, 1997.