Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

[SOLVED] OFDM with no carrier

Status
Not open for further replies.

warriorwithin

Newbie level 4
Joined
Jan 12, 2012
Messages
7
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Activity points
1,325
My system uses 0to2khz freqs. Hence I do not want to use any RF carrier.Is it possible to generate OFDM signal in this BW range and transmit it without using any carrier. If so what factors determine the carrier freqs and no of carriers. My BW and Symbol time is known and fixed.

Thanks
 

No you cannot do this.
Simplifying very much, the efficiency of the antenna, that is the power transmitted over the air with respect to that sent to the antenna is proportional to the frequency , approaching zero going towards the DC (it is not possible to transmit a DC voltage). This means you must have a gain (at RX or TX or both sides) increasing as 1/f, that is infinite at DC
 

You can transmit your signal with no carrier by cable channel.

If you have symbol time T then ofdm subcarrier spacing is determined as 1/T. Number of subcarriers is BW*T.
But some of that sucarriers you should assign to cyclic prefix so the number of subcarriers assigned for user data will be a little bit less that total.
 

Thanks Mityan and albbg for your reply.

@Mityan

Usually they multiply the I and Q with Fc with 90deg phase diff and then add them together and transmit. How will the circuit be in this case when Carrier is not used.
 

I think the same as for traditional OFDM excluding one thing: the symbol corresponding to 0's subcarrier (0 frequency) should be dummy because for constant 'phase' term is not applicable.
The algorithm may be like this:
1. For your data bit flow you collect M bits according to your symbol constellation dimension - e.g. 2 bits for qpsk, 4 bits for 16qam etc.
2. Map to symbol amplitude-phase, converting them to complex numbers
3. For FFT of length N collect N-1 symbols assuming symbol #0 is dummy (0).
4. Perform the FFT (it means you shift every symbol to its subcarrier from 1 to N-1 within the given whole band).
5. Then take real parts of FFT result and output to data communication channel (because only real signal goes over the channal always, complex is just useful representation for processing).
You will get a number of subcarriers spaced by 1/T (from 1/T to (N-1)/T) with amplitudes and phases according to their input symbols.
For time T you should have M*(N-1) bits - this will define data rate and band.
For real application you also should provide a guard interval/cyclic prefix. And I'm not sure about syncronization.
 

5. Then take real parts of FFT result and output to data communication channel (because only real signal goes over the channal always, complex is just useful representation for processing).
You will get a number of subcarriers spaced by 1/T (from 1/T to (N-1)/T) with amplitudes and phases according to their input symbols.
For time T you should have M*(N-1) bits - this will define data rate and band.
For real application you also should provide a guard interval/cyclic prefix. And I'm not sure about syncronization.

IF I am taking only the real parts of IFFT and outputing to the channel, then what do I input to the FFT at reciever? The FFT requires both I and Q signals as input, isn't it?
 

No, it is not. There is no requirement to FFT input function to be a complex array.
 

No, it is not. There is no requirement to FFT input function to be a complex array.

Hi

I am unable to get back my original sequence by taking FFT of only the real part. I think there must be some pre/post processing required.
Could you please help me out in this
 

I am unable to get back my original sequence by taking FFT of only the real part. I think there must be some pre/post processing required.
Could you please help me out in this
Over radiochannel data is transmitted by variation of amplitude/phase of carrier. The first is square root of sum of IQ components squared, the second is arctangent of Q/I. So maybe real channel data will be sum of I and Q like it is in all IQ modulator diagrams?
 

Came to know in this case that we need to take 2N pt IFFT with the other half of inputs being the conj symm extention of 1st half. This kind of IFFT gives rise to only real valued signal which can be transmitted over a line. (u get a 2N real signal, we need to ignore the 2nd half)
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top