**- DSP log - http://www.dsplog.com -**

Negative Frequency

Posted By __Krishna Sankar__ On August 8, 2008 @ 6:05 am In __DSP__ | __25 Comments__

Last week, I received an email from Mr. Kishore. He was wondering about the physical significance of **negative frequency**. Does **negative frequency** really exist?

Though I have seen conflicting views on the net (thread in complextoreal.com ^{[1]}, thread in comp.dsp ^{[2]}), my perspective is that **negative frequency exist**. The concept of negative frequency helps me a lot to understand single sideband modulation (SSB), OFDM systems, I Q modulators etc (to name a few).

The wiki entry on negative frequency ^{[3]} provides a simple explanation using as an example.

We know that . This means the sign of cannot be un-ambiguously found out from observing alone. This implies that it is reasonable to think that has frequency components at both and .

Similarly, this ambiguity exists for too.

Thanks to the nice paper by Mr. Richard Lyons, Quadrature Signals: Complex but not Complicated, Richard Lyons ^{[4]}.

Let us first define the magic number and so on.

The Taylor series expansion of , and are as follows.

Let us now define the Taylor series expansion of .

.

Similarly,

.

Combining the above two equations, one may write

.

If we apply , then we get,

and

.

This forms the proof that a real sinusoidal having frequency is comprised of a complex sinusoidal having a **positive frequency** at and a **negative frequency** at .

Click here to download Matlab/Octave code for plotting the spectrum of real and complex sinusoidal ^{[5]}

**Figure: Spectrum plot showing positive and negative frequency**

As discussed above, the real sinusoidal has frequency components at +5MHz and -5MHz where as the complex sinusoidal has frequency component only at +5MHz.

Some of you might be familiar with the IEEE 802.11a specification ^{[6]} where subcarriers from [-26 to -1 and [+1 to 26] are used. The subcarriers -26 to -1 corresponds to usage of negative frequency and lets try to understand it.

The equation for an OFDM transmission is,

,

where

(a) correspond to the frequency of the sinusoidal and

(b) is a rectangular window over

(c) is the symbol period.

(d) each information signal is modulated on to a **complex sinusoidal** having frequency of .

(e) Sum of all such modulated sinusoidals are added and the resultant signal is sent out as .

In the IEEE 802.11a specification ^{[6]}, symbol duration is 3.2, sampling frequency is 20MHz and .

The frequencies used for modulating the ‘s are , , , and so on till . Expressing in Hz, this corresponds to frequencies from 0Hz, 312.5kHz, 625kHz, 937.5kHz, 1.25MHz,…, 10MHz, 10.3125MHz,… till 19.6875MHz.

From our understanding of sampling theory, we know that with a sampling frequency of , we can only see frequencies from to .

Note: The frequency is called the Nyquist frequency. ^{[7]}

So in our 802.11a example, what will happen to frequencies which are modulated on subcarriers lying from 10MHz till 19.6875MHz?

**Quick answer**: They get folded!

The frequencies from **10MHz till 19.6875MHz** gets folded and seems as if they are lying from **-10MHz to -312.5kHz**.

**Figure: Spectrum folding to negative frequency in IEEE802.11a specification**

Further, folding of the spectrum to the **negative frequency region does not cause any problems**. Reason: ‘s which where modulated on complex sinusoidals having frequencies from 0 till 10MHz **did not have any negative frequenc**y component.

**Note:**

The concept of folding is well explained in Chapter 1.4.1 of [DSP: PROAKIS] ^{[8]}. A simple example of folding which we may see in our day to day life is with a ceiling fan. One may see that the blades of the fan are rotating at a slower speed in a direction opposite to the actual rotation of the fan blades. Needless to say that the sampling frequency of our eye’s are not good enough.

Happy learning.

[DSP: PROAKIS] ^{[8]}: Digital Signal Processing, John G. Proakis ^{[8]}

Quadrature Signals: Complex but not Complicated, Richard Lyons ^{[4]}

Article printed from DSP log: **http://www.dsplog.com**

URL to article: **http://www.dsplog.com/2008/08/08/negative-frequency/**

URLs in this post:

[1] thread in complextoreal.com: **http://complex2real.proboards18.com/index.cgi?board=commq&action=display&thread=63**

[2] thread in comp.dsp: **http://groups.google.com/group/comp.dsp/browse_thread/thread/ed84c50cb5245959/5b65bcb16a30f336?q=negative+frequency&lnk=ol&**

[3] wiki entry on negative frequency: **http://en.wikipedia.org/wiki/Negative_frequency**

[4] Quadrature Signals: Complex but not Complicated, Richard Lyons: **http://www.dspguru.com/info/tutor/QuadSignals.pdf**

[5] Matlab/Octave code for plotting the spectrum of real and complex sinusoidal: **http://www.dsplog.com/db-install/wp-content/uploads/2008/08/negative_frequency.m**

[6] IEEE 802.11a specification: **http://standards.ieee.org/getieee802/download/802.11a-1999.pdf**

[7] Nyquist frequency.: **http://en.wikipedia.org/wiki/Aliasing**

[8] [DSP: PROAKIS]: **http://www.amazon.com/gp/redirect.html?ie=UTF8&location=http%3A%2F%2Fwww.amazon.com%2FDigital-Signal-Processing-John-Proakis%2Fdp%2F0131873741&tag=dl04-20&linkCode=ur2&camp=1789&creative=9325**

[9] click here to SUBSCRIBE : **http://www.feedburner.com/fb/a/emailverifySubmit?feedId=1348583&loc=en_US**

Click here to print.

Copyright © 2007-2012 dspLog.com. All rights reserved. This article may not be reused in any fashion without written permission from http://www.dspLog.com.