Jim:
If I may add my two cents...
The question being begged is, " Given equally efficient multiple access protocols, which modulation scheme is more spectrally efficient?"
I think the question must be begged.
And of course, the response would be: " The two are indistinguishable; you cannot talk about one without the other."
Yep.
Is there no way to resolve this question, even on a hypothetical basis?
IMHO, No.
Is there no cell phone modulation scheme, existing or potential, about which you would say "Purely on the basis of spectral efficiency, I would vote for this standard?"
The simple answer is no, because, again IMO, they are all equally efficient or not efficient in a fully loaded system, and when performing an "apples to apples" comparison. Unfortunately, it is very difficult to do apples to apples comparisons. When you adjust the parameters of the test to make a fair comparison, each proponent howls that by doing so, you diminish or negate one of their key advantages.
Take a quick look at the CDMA / TDMA debate.
Qualcomm is credited, and rightfully so, with 3 particular innovations, precise power control, rake receivers, and soft handoff . The first made CDMA in a terrestrial mobile environment possible, the second mitigates the effects of multipath and the third diminished the incidents of dropped calls. Only the third affects system capacity, and deleteriously so.
Using the simplest math, Digital AMPS supposedly gets 3x analog amps capacity. CDMA proponents have claimed 6-10x depending on which vocoder being used. Actual implementations are generally lower. Qualcomm must have done something to increase spectral efficiency right?
Lets give CDMA the benefit of the doubt and assume 10x analog AMPS capacity. Well, half of the capacity gain is due to a feature - voice activity detection - which is NOT present in the digital AMPS standard. Basically, the transmitter turns off when there is no conversation. No transmit, no interference. Take away this feature, and the upper bound of CDMA capacity is 5 the lower bound 3. The lower bound looks vaguely familiar: isn't that the Digital AMPS / TDMA multiple? Now, Digital AMPS being an older standard just did not have this feature in it - GSM revisions have included it. If one attempts to compare apples to apples - and I'm ignoring a lot here for simplicity - the capacities are strikingly similar. Of course CDMA proponents argue that you have to compare what they offer to the alternatives - and they are right. But that doesn't mean that if one is designing from scratch - without any legacy equipment or protocols one can't do just as well. (Furthermore, I am neglecting here the negative effects of CDMA soft handoff on capacity and the capacity gains that a competent RF designer can get out of TDMA using overlay / underlay techniques, non-standard frequency plans etc., The gulf is just not that wide, and some would argue its more likened to a gully - if that.)
Basically what you gain in efficiency - as measured in bits per hertz - is subtracted from range. Why? As a general rule, as you increase efficiency, you must maintain a higher signal to noise ratio.
Lets look at that fully loaded system part. Every transmission must maintain a certain minimum signal to noise ratio in order to decode the signal. Basically, CDMA is an interference averaging technique. Interference is noise, by another name. Whether the ether is filled with noise from a multitude of TDMA or CDMA users - its still filled! The S/N ratio is violated, errors commence, the transmission fails.
Let me try an analogy.
Imagine the ether as an rectangular aquarium. The bottom of the aquarium is the noise "floor". The height of the aquarium represents an absolute distance or range, i.e., the distance the signal can travel without error. That height, the brim of the aquarium is the point where the minimum S/N ratio is violated. (Remember, interference is noise by a different name.)
In that aquarium one places infinitely thin, square glasses each filled to its brim with water. The brim of these glasses exactly matches the lip of the aquarium. One fills this aquarium with my infinitely thin walled glasses until there is no free space.
If I place another glass on top or pour its contents in, the aquarium overflows, i.e, the minimum S/N is penetrated.
Assume, my infinitely thin-walled glasses were TDMA transmissions, the water in those glasses effectively the noise/interference each contributed to the ether. Let me extend it to CDMA.
CDMA transmission are not contained in time, (glasses in the analogy) they just enter the either. I start pouring the contents of the glasses into the aquarium. Regardless of the number of infinitely thin-walled glasses I placed into the aquarium, the contents of those exact same number of glasses still fills the aquarium to the brim. Pour one more glass in, the aquarium overflows.
I would argue - in the abstract - that no technique is better than another in a fully loaded system because the ether can only contain so much energy, before it becomes impossible to decode a signal. Again, using my analogy, you can have a technique which gets more bits per hertz, but then the glasses need to be bigger - more bits / more water. One fills the aquarium faster. Again, if we liken the brim of the aquarium to a distance the signal can travel before the signal to noise ratio is violated, then it becomes obvious, you don't get something for nothing.
But what about....
All, IMHO.
ww |
