please help in to understand basic of cyclic prefix

[the_emperor_966]

hello , i have a problem to understand how the cyclic prefix change the linear convolution into circular convolution , i know that that we copy the last symbols and paste it in the beginning of the symbols just i want to understand if we put for example ones instead of the cyclic prefix what is the different ??

 

[mmoctar]

Because fft is cyclic, if you got N symbol the 0 and the N+1 symbol is the same that why when you put the end of your ofdm symbol at the begin the say it cyclic prefix. But the are some applications who use zero padding for using few power but they lose orthogonality more than when you use prefix cyclic.

 

[rag_perfect]

i have also one doubt...
as per theory... by adding CP we are making OFDM symbol periodic so that we can apply the property:Convolution in time doman= multiplication in frequency domain in Channel Equazation part ,which is valid only for Circular convulation.

but if you consider multiple OFDM symbols; add individual CP in all of them and combine them then composite Symbol(as called 'Frame' in WimaX case) will not look like Periodic signal anyway??? That means purpose of Circular convolution fails here!!
plz explain????

 

[the_emperor_966]

Re: please help in to understand basic of cp ( please again

is i am with you but if we put zeros instead of cp it is work without any problem (except the power) i mean work correctly in the system , and if we put for example just ones instead of cp and in the receiver we deleted from the symbol it is correclty work, and ihave another question iead here in one of the replays in edaboard that's he make the length of the cp smaller then the impulse response of the channel (but i know that the cp must be longer than the length of the impulse response of the channel ) explain why ????
this the replay i saw



thank for you all.....

 

[rag_perfect]

hi the_emperor_966..
The CP should always greater than channel impulse respose and in reply also he said exactly this.probably you have noticed it wrong...
nways...thankx 2 the link u provided ..it clerifies my doubt also...

however.. plz see this link on dsprelated.. it clerifies more regarding use of CP and Zero prefix......

**broken link removed**

 

[the_emperor_966]

hi ... i have ap problem to understand the cp in ofdm, i did this program matlab upon my information
x=[0 1 2 3 4 5 6 7 ];
>> h=[0 1 2 ];
>> xx=ifft(x);
>> xx=[xx(7:8) xx]

xx =

Columns 1 through 3

-0.5000 + 0.5000i -0.5000 + 1.2071i 3.5000

Columns 4 through 6

-0.5000 - 1.2071i -0.5000 - 0.5000i -0.5000 - 0.2071i

Columns 7 through 9

-0.5000 -0.5000 + 0.2071i -0.5000 + 0.5000i

Column 10

-0.5000 + 1.2071i

>> y=conv(x,h);
>> fft

ans =

Columns 1 through 3

84.0000 -14.8541 +40.2874i -11.8541 +11.1352i

Columns 4 through 6

-8.1459 + 1.3876i -5.1459 - 1.0041i -4.0000

Columns 7 through 9

-5.1459 + 1.0041i -8.1459 - 1.3876i -11.8541 -11.1352i

Column 10

-14.8541 -40.2874i

here y doesn't equal to x ( but i know that it's must be the same)

please....... what is my problem????

thank you............

 

[shankarrajv]

hi
i want to discuss about the cyclic prefix in ofdm. cyclic prefix are used bcos when transmitting the msg due to multipath interference the signal which was sent (n-l)before and the current signal n get interfered i.e) the tail of n-1 symbol will interfere with the head of nth signal. the delay(the length of interfernce depends upon the impulse respose of the channel) so if we choose the length of CP (>impulse response)in such a way the interference will not affect the remaining data. As evenit get affected due to cyclic property we can get the data as it is