FDMA vs. CDMA

Fundamentals
Title	FDMA vs. CDMA
Author(s)	J.A Ávila Rodríguez, University FAF Munich, Germany.
Level	Intermediate
Year of Publication	2011

GPS, Galileo and Compass are or will be using CDMA while GLONASS is the only one that still employs FDMA for the transmission of its navigation signals. However, the Russian Navigation System is moving in the direction of achieving higher interoperability with the American, European and Chinese systems as GLONASS has already taken the first steps to CDMA.

In spite of the promising benefits that this change would bring, the main reasons that the Russian GLONASS has used as argument in the past against CDMA are summarized next:

• Existence of single point of failure if all signals are located at E1/L1,
• FDMA offers improved security protection (not any more true as we explain next),
• The issue of paying for the new civil signal design,
• The historical reasons that lead to FDMA

The improved robustness of FDMA versus CDMA has usually been justified by the improved Spectral Separation Coefficient (SSC) that FDMA can achieve. If we take as an example the C/A Codes of GPS and GLONASS, we can see that the Self SSC of the GLONASS C/A Code is of approximately -57.9700 dB-Hz while for GPS we obtain -61.8008 dB-Hz. The favourable difference for the GPS C/A Code comes from the different employed transmission filters and the lower code rate of the GLONASS C/A Code. However, if we take a look at the SSC between spectral adjacent GLONASS C/A Codes, we can recognize than the spectral separation improves to -69.5604 dB-Hz, providing thus nearly 12 dB of additional protection with respect to CDMA. Moreover, if we take a look at non-adjacent spectra the theoretical isolation is infinite and thus the average of the adjacent SSCs would be of -80.6999 dB-Hz or nearly 22 dB better (considering 14 frequency slots).

CDMA achieves higher protection by means of the cross-correlation of the employed codes. These provide in the case of the GPS C/A Gold Codes an additional protection of 24 dB when no Doppler is considered and of approximately 21 dB when also Doppler is taken into account. In the case of FDMA there is only one code and thus we have to talk about auto-correlation instead of cross-correlation. Nonetheless, although all the GLONASS satellites employ the same code for all the satellites, since the relative Doppler and delay among them can be considered as random, the final autocorrelation value that two FDMA satellites present is that of the secondary peaks of the ACF, also in the order of 21 dB with respect to the main peak, and thus close to the cross-correlation of GPS C/A Codes. The randomization effect through Doppler is similar to the principle of Doppler Division Multiple Access (DDMA).

If we consider now the spectral separation and code separation effects together, we can see that from this point of view FDMA would provide an additional protection of 22 dB with respect to the CDMA approach. This has been indeed the main argument used by GLONASS experts until now. As we can recognize, the implicit assumption behind is that the jammers are narrowband.

While the protection against jammers was at the beginning of GNSS of particular importance and drove most of the decisions that Russia and the USA took to build their respective satellite navigation systems, nowadays it is possible to build very wideband jammers. Thus, the protection that FDMA was supposed to offer against narrowband interferers is not any more an advantage against CDMA.

We can imagine it with a simple example. If we jam a satellite in a CDMA system with a narrowband interferer, we automatically jam all the other satellites since they are on the same carrier. On the contrary, in an FDMA system the other satellites would result in principle unaffected. Wideband jammers are however not an issue or at least not as they were at the beginning when GLONASS argued the FDMA goodness on the basis of its superior jamming protection. As a result, unless the different satellites used carriers very separated in frequency, with today’s technology one could jam all the FDMA signals at the same time disabling the extra protection that FDMA was supposed to bring.

In addition, FDMA is a clear show-stopper for mass market applications since having different carriers for each satellite poses an important challenge in the design of the receivers. As one can imagine, this makes FDMA less competitive than its CDMA competitor. Moreover, filter design and other synchronization aspects difficult the design of an FDMA receiver. Although the problem can be solved as many manufacturers have shown in the past years, there is no doubt that if GLONASS wants to really provide mass-market signals, its civil signals shall slowly migrate to CDMA.