Equalization for FH-4FSK system

Hi, I would like to design use an equalizer for improving the performance
of 4fsk-frequency hopping system in rayleigh fading, I ve seen matlab
functions but they all are for linear modulations or is there any function
for non linear modulations??. Can any one please guide me on this issue? Is
there any particular technique that is useful for this case?? thanks

On Feb 5, 3:41 pm, "furqee" <fur...@ieee.org> wrote:
> Hi, I would like to design use an equalizer for improving the performance
> of 4fsk-frequency hopping system in rayleigh fading, I ve seen matlab
> functions but they all are for linear modulations or is there any function
> for non linear modulations??. Can any one please guide me on this issue? Is
> there any particular technique that is useful for this case?? thanks

Take a look at the Constant Modulus Algorithm (CMA).

John

furqee wrote:

> Hi, I would like to design use an equalizer for improving the performance
> of 4fsk-frequency hopping system in rayleigh fading,

The whole point of the frequency hopping is to decorrelate the fading. 
Hence every hop has to be equalized independently of the other hops.

> I ve seen matlab
> functions but they all are for linear modulations or is there any function
> for non linear modulations??.

Despite of the popular belief, Matlab is not a replacement for the brains.

> Can any one please guide me on this issue? Is
> there any particular technique that is useful for this case?? thanks

Traditional equalizers are not very usefull here because the 
equalization has to be performed for each hop. The simplest algorithm 
would be running the LMS by several iterations through the hop; there 
could be much better solutions depending on the particulars.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

Thanks for replying,John and Vladimir, I have just seen few IEEE papers on
the equalization of nonlinear modulations like CPM,FSK etc.
What I intend to do in matlab now is to first dehop FH signal to baseband
so that its just complex vector having 4 tones of quaternary CPFSK, with
continous phase in each hop and then apply equalizer on each hop duration,
would you kindly tell me whether its right?
 and is there any difference between equalizing a simple CP-4FSK signal
and equalizing a FH-4FSK after dehopping other than we have to apply
equalizer to each hop?? 


>
>
>furqee wrote:
>
>> Hi, I would like to design use an equalizer for improving the
performance
>> of 4fsk-frequency hopping system in rayleigh fading,
>
>The whole point of the frequency hopping is to decorrelate the fading. 
>Hence every hop has to be equalized independently of the other hops.
>
>> I ve seen matlab
>> functions but they all are for linear modulations or is there any
function
>> for non linear modulations??.
>
>Despite of the popular belief, Matlab is not a replacement for the
brains.
>
>> Can any one please guide me on this issue? Is
>> there any particular technique that is useful for this case?? thanks
>
>Traditional equalizers are not very usefull here because the 
>equalization has to be performed for each hop. The simplest algorithm 
>would be running the LMS by several iterations through the hop; there 
>could be much better solutions depending on the particulars.
>
>
>Vladimir Vassilevsky
>DSP and Mixed Signal Design Consultant
>http://www.abvolt.com
>
>
>

On Feb 5, 12:41 pm, "furqee" <fur...@ieee.org> wrote:
> Hi, I would like to design use an equalizer for improving the performance
> of 4fsk-frequency hopping system in rayleigh fading, I ve seen matlab
> functions but they all are for linear modulations or is there any function
> for non linear modulations??. Can any one please guide me on this issue? Is
> there any particular technique that is useful for this case?? thanks

If the 4fsk system is tightly constrained by bandwidth (for maximal
spectral efficiency), then a MLSD works well. An MLSD for non-coherent
detection will not have the performance of a coherent receiver but
will still be beneficial and a lot less complex. This was what was
used in Proxim's 10 Mbit WLAN frequency hopping system (homerf 2.0).
Without knowing the modulation index (h) of the 4fsk, though it is
hard to generalise