OFDMA vs. SC-FDMA

[ffyj1901]

Greeting,

I have following questions to discuss. They are listed below. Thanks in advance.

1. In 4G or LTE standard, OFDMA is used for downlink and SC-FDMA is used for uplink. SC-FDMA is supposed to be a modification of OFDM ( with extra DFT and frequency mapping block and everything else stays the same). How do the two extra added blocks ( DFT and freqeuncy mapping) make SC-FDMA a single carrier modulation scheme?

2. The main reason using OFDM is for its fading performance. It is due to the long symbol duration because the signal is converted from serial to parallel. For SC-FDMA, the signal is transimitted on one carrier, can SC-FDMA really be called a OFDM system? How does SC-FDMA maintain the fading performance with shorter symbol duration and single carrrier?

3. OFDM is primarily for indoor application. With LOS, the crosstalk will degrade the OFDM performance. Is OFDM really suitable for cellular application?

4. Is HSPA (3.5G) based on W-CDMA? Are the difference between 3G and 3.5G mainly due to the modulation scheme? Thanks for your time and look forward to hear your feedback.

Sincerely,

Louis

[charan langton]

Answers.

1. In 4G or LTE standard, OFDMA is used for downlink and SC-FDMA is used for uplink. SC-FDMA is supposed to be a modification of OFDM ( with extra DFT and frequency mapping block and everything else stays the same). How do the two extra added blocks ( DFT and freqeuncy mapping) make SC-FDMA a single carrier modulation scheme?

What makes SC-FDMA a single carrier modulation is the fact that that there is just one carrier. The FDMA part can have many other carriers and unlike OFDM, they are usually not orthogonal. For exmaple, you can have carriers that are 22 MHz apart as in satellite transponders.

2. The main reason using OFDM is for its fading performance. It is due to the long symbol duration because the signal is converted from serial to parallel. For SC-FDMA, the signal is transimitted on one carrier, can SC-FDMA really be called a OFDM system? How does SC-FDMA maintain the fading performance with shorter symbol duration and single carrrier?

The fading protection comes from the fact that the fading is frequency selective and will effect just one symbol at a time. Also the cyclic prefix helps as well to remove ISI.
SC-FDMA came first, historically speaking. No, I dont think anyone calls SC-FDMA a OFDM system.
The more important reason is that in OFDM, the carriers are orthogonal and in SC-FDMA that is not a requirement. They are usually not othogonal.

3. OFDM is primarily for indoor application. With LOS, the crosstalk will degrade the OFDM performance. Is OFDM really suitable for cellular application?

OFDM is used outside as well. In LOS, OFDM should perform pretty well but that would be a waste of power. MPSK modulation works most efficiently in LOS. Cross talk is no more a problem for OFDM than QPSK. Sirius radio ground links are OFDM. OFDM requires much higher Eb/N0 but it does offer fading protection and in some cases, mostly in broadcast, this is a good choice.

4. Is HSPA (3.5G) based on W-CDMA? Are the difference between 3G and 3.5G mainly due to the modulation scheme? Thanks for your time and look forward to hear your feedback.

I can only answer this question on the waveform level. I am sure there are many other changes in 3.5G. The waveform is adaptive and can change from QPSK to 16QAM. Not sure about the CDMA part. I dont think it is.

Charan Langton

Louis [/quote]

[ffyj1901]

Dear Charan:

I still think SC-FDMA is derived from OFDM. Here is the reason why. Please see the AN from Agilent at cp.literature.agilent.com/litweb/pdf/5989-8139EN.pdf. Looking at the hardware implementation in figure 11 and comparison in figure 8, SC-FDMA used in LTE system has everything OFDM has. In addition to the standard OFDM, SC_FDMA has the DFT and data mapping block. I think it is where I am unclear on this topic. If SC-FDMA is unrelated to OFDM, why are the DFT and IFFT block needed? By using IFFT, you essentially add multiple carriers together to form the final waveform, so IFFT will make SC-OFDM a multi-carrier system. That goes back to my first question why SC-FDMA is a single carrier system. Please advise.

Regarding to why SC-FDMA still keep the good fading performance in spite of single carrier, Agilent's explanation is on the first the paragraph on page 20. They are saying it is due to "constant nature of each sub-carrier" which doesn't make sense to me. Please advise as well.

Thanks in advance for your time and your help.

Sincerely,

Louis

[hylin]

Hello Louis,
I think the SC-FDMA is Single-Carrier, not Single Frequency. The data signal of each user consists of a lot of frequecny. DFT of SC-FDMA is used to filter the frequecny items and map them into IDFT to refrom single user waveform. Unlike OFDMA, IDFT consists of the singal from a lot of different users. It leads to large PAPR. Hope that this is helpful.
BR,
HY Lin

[charan langton]

Understanding SC-FDMA as used in LTE

Let's assume that there are 4 mobiles talking to one base. If we use a OFDM signal for each one, then the max PAR is equal to IFFT size, assume 64.

Now tell each mobile to confine its signal to the first 16 carriers and the second one to the second 16 carriers, so on. How can a mobile do that? If it does a 64 point IFFT, the signal will be spread over all 64 carriers.

The way is for it to do a 16 point DFT. Now the data is spread over just 16 carriers. (time to frequency) Each mobile does the same thing, has a 16 carrier signal.

Now by pre-arrangement, each mobile is told just exactly where it will array its carriers in the 64 carrier OFDMA space (note the A at the end, which makes OFDM a Multiple Access system.) The first mobile now does a 64 point IFFT by first mapping its data to just the first 16 positions, and filling the rest with zeros, also called zero-padding.

The second mobile does the same, except it puts its data in the second block, pads the rest with zeros and does a 64 point IFFT.

Each mobile sees a max PAR of only 16, instead of 64. This is a big reduction.

The outgoing signal for each mobile after having done a 16 point DFT and then a zero-padded 64 point IFFT is a single-carrier signal. What is the impact of the larger size IFFT? The larger IFFT only adds some time-domain interpolation but other than that this is the same data at the input. Now it is transmitted at a single assigned frequency.

When four such signals are received by the base even if at the same frequency, the base can extract out the individual signals. It first does one 64 point FFT (time to frequency) and sees carriers from all four mobiles. It then picks out each of the blocks, first 16 for mobile 1..., and does a IDFT to convert the signal back to time domain.

The mapping can be distributed instead of block-based. Some are called LFDMA, MFDMA etc.

Charan Langton

[kmouz]

Greetings,

Regarding the difference between 3G and 3G+, it is not only the 16QAM modulation, it is also the fact that in 3G dedicated channels are used while in 3G+ in the donwlink there is a high speed common channel that is shared by all users.

With 3G++ or (HSPA+), 64QAM modulation, MIMO and Dual-Cell techniques are used to increase the peak data rate.

I hope this helps.

Regards
Khalil Mouzawak