Coverage comparison for wireless systems at various frequencies

One of the items I've been meaning to add to my wireless tutorial is a discussion of actual coverage areas for different frequency bands.  This matters because, with today's mobile radio technology, lower frequencies cover more distance and do better at penetrating buildings.  That means fewer cell sites for equivalent coverage and thus lower infrastructure costs.

Indeed, a little discussed issue in the US is that Verizon and AT&T own the 850 MHz cellular spectrum and they were the big winners in the recent 700 MHz auctions.  Sprint PCS and T-Mobile USA use spectrum at 1900 MHz or 1700 MHz, so they need more cell sites to achieve similar coverage.  They are at a cost disadvantage.

Signals go even further using the 450 MHz spectrum that's available in some countries.  I've argued publicly that the best thing developing countries can do to bring mobile coverage to rural areas (for example in India), or to remote areas in general, is to make 450 MHz spectrum available to mobile operators.

To get a sense of how significant these effects are, here's a table that Qualcomm submitted to the ITU's Working Party 8F several years ago.  (Thanks to Joe Nordgaard for the pointer).

Frequency (MHz)	Cell radius (Km)	Cell area (Km2)	Relative cell count
450	48.9	7521	1
850	29.4	2712	2
950	26.9	2269	3.3
1800	14	618	12.2
1900	13.3	553	13.6
2500	10	312	24.1

Source: Qualcomm ITU 8/F Submission, June 11, 2001, “Coverage comparison of systems at various frequency ranges, including 450 MHz”

Comments

[Harmeet]

Hi,

I have the following design problem..

Hello,

I have the following design problem...

Assume the FCC has opened a band of spectrum for commercial use. The spectrum is centered around 3 GHz. Your company has decided to use this spectrum to deploy a wireless voice telephony service to
compete with the existing cell phone services. Your boss in this imaginary company has charged you with working out possible system designs, and deciding which to use. However, the spectrum has some strange constraints. Transmit power levels are limited to 2 Watts per RF carrier (so a base station could transmit more than this since it might have more RF carrier frequencies). The TOTAL bandwith available in this spectrum band is 2 MHz.

No other constraints are there. We can be fairly liberal in our assumptions.. Now i need to know what things that I can research to make this as detailed as possible. It is not going to be a thesis its just a design problem and I am required to write a paper listing all the different thing in which the goal can be achieved. I think the major part will be to determine which system design (CDMA, TDMA etc ) will be better suited for the problem.. kindly let me know any comments have on this.

[Brough Turner]

Harmeet, I can't offer to do your homework for you. :-) But let me point out a gap in your specification and then throw out some thoughts you might investigate...

First, you say 2 watts per RF carrier but then don't specify the bandwidth per carrier. Typically, power limits are specified in watts per Hz, KHz or MHz. If it's watts per carrier you have to specify the carrier width.

Other thoughts: You will have a hard time competing with 3G cellular operators with only 2 MHz total spectrum. The cellular operators typically have much more spectrum -- minimum 10 MHz for 3G UMTS. So competitively, you will need a lot of cell sites to achieve comparable capacity, not to mention needing more cell sites in order to achieve full coverage at 3 GHz.

For a design-from-scratch, the most efficient current technology is based on OFDMA and variations. Look at what's used in WiMAX and LTE. They are very similar and state-of-the-art. Including multiple input multiple output (MIMO) technology and that's likely to be the best you'll do with current semiconductor technology.

If you want to go out on a limb, it might be interesting to consider combining full power use of the permitted 2 MHz of bandwidth with ultra wideband transmissions below the noise floor of everyone else between 3 GHz and 10 GHz (or perhaps from DC to ultra-violet - just kidding...).