IIR filter question

Hi Guys,
   I was asked to review a potential article 
for the IEEE Sig. Proc. magazine.  In that 
article the author implies that IIR (recursive) 
filters aren't as popular nowadays as they were in 
the past (say 10-20 years ago).  

Now I'm no IIR filter designer, so I need your 
opinions.  Aren't IIR filters still as popular 
now for audio signal processing as they were 
10 years ago?

Thanks guys,
[-Rick-]

Hello Rick,


>    I was asked to review a potential article 
> for the IEEE Sig. Proc. magazine.  In that 
> article the author implies that IIR (recursive) 
> filters aren't as popular nowadays as they were in 
> the past (say 10-20 years ago).  
> 
> Now I'm no IIR filter designer, so I need your 
> opinions.  Aren't IIR filters still as popular 
> now for audio signal processing as they were 
> 10 years ago?
> 

I am no expert either but from what I have seen IIR is still used, often 
with some added measures to muffle their inherent instability upon 
signal loss.

For narrowband applications wave digital filters (WDF) have made inroads 
and probably eaten away at IIR's "market share", especially in the 
telecom business. The fact that such WDF can often live with less 
computational resources might not matter much anymore in the audio world 
but it does if you have to do filtering at high clock rates. In the 
passband they are not very sensitive to coefficient granularity so you 
can skimp on the number of barrel-shift and add functions. This cuts 
down real estate on a chip.

Regards, Joerg

http://www.analogconsultants.com

R.Lyons@_BOGUS_ieee.org (Rick Lyons) writes:

> Hi Guys,
>    I was asked to review a potential article 
> for the IEEE Sig. Proc. magazine.  In that 
> article the author implies that IIR (recursive) 
> filters aren't as popular nowadays as they were in 
> the past (say 10-20 years ago).  
>
> Now I'm no IIR filter designer, so I need your 
> opinions.  Aren't IIR filters still as popular 
> now for audio signal processing as they were 
> 10 years ago?

I think the author(s) may be right. The situation is
that we have much more powerful processors now than
we did 10-20 years ago, so we can use brute-force 
FIRs instead of IIRs. I don't think there's as big
of a need for IIRs as there was in the past. Kinda
like the way you don't see a lot of assembly language
nowadays, either.
-- 
%  Randy Yates                  % "I met someone who looks alot like you,
%% Fuquay-Varina, NC            %             she does the things you do, 
%%% 919-577-9882                %                     but she is an IBM."
%%%% <yates@ieee.org>           %        'Yours Truly, 2095', *Time*, ELO   
http://home.earthlink.net/~yatescr

Hi Rick,

Rick Lyons wrote:
> Hi Guys,
>    I was asked to review a potential article 
> for the IEEE Sig. Proc. magazine.  In that 
> article the author implies that IIR (recursive) 
> filters aren't as popular nowadays as they were in 
> the past (say 10-20 years ago).  
> 
> Now I'm no IIR filter designer, so I need your 
> opinions.  Aren't IIR filters still as popular 
> now for audio signal processing as they were 
> 10 years ago?
> 
> Thanks guys,
> [-Rick-]
> 

I think this is a difficult call, since so many applications exist and 
any one of us only sees a small portion of them. My feeling is IIRs are 
still used where they have always been used, but most new areas seem to 
use FIRs these days.

I guess noise shaping has kept IIRs entrenched in some areas they might 
otherwise have been kicked out from. There are still plenty of brand new 
audio things using IIRs with noise shaping. This is especially true in 
cheap consumer equipment where "fully digital" sells, but prices are 
stripped to the bone.

Two of the people to whom I've recommended your book seem to feel IIRs 
are obsolete after reading it. :-)

Regards,
Steve

Randy Yates wrote:
> R.Lyons@_BOGUS_ieee.org (Rick Lyons) writes:
> 
>> Hi Guys,
>>    I was asked to review a potential article 
>> for the IEEE Sig. Proc. magazine.  In that 
>> article the author implies that IIR (recursive) 
>> filters aren't as popular nowadays as they were in 
>> the past (say 10-20 years ago).  
>>
>> Now I'm no IIR filter designer, so I need your 
>> opinions.  Aren't IIR filters still as popular 
>> now for audio signal processing as they were 
>> 10 years ago?
> 
> I think the author(s) may be right. The situation is
> that we have much more powerful processors now than
> we did 10-20 years ago, so we can use brute-force 
> FIRs instead of IIRs. I don't think there's as big
> of a need for IIRs as there was in the past. Kinda
> like the way you don't see a lot of assembly language
> nowadays, either.

Now don't get me started on horrible zero-based assembly language, Randy...

Actually, I disagree.  The reason is that, in my experience, IIR filters 
are pretty-much impractical in fixed-point processors because they 
typically have to be implemented using multiple words of precision.  For 
that reason, with the exception of very simple DC blocking filters, I 
have only once used IIR filters in a fixed-point application--and then 
only because it was extremely memory starved.  (In case anybody's 
wondering, that's pretty-much why Iowegian doesn't sell an IIR filter 
design product yet: I've never needed such a thing myself.)

However, IIR filters are practical in floating-point processors. 
Therefore, given the availability of fairly cheap floating point 
processors as of about 10 years ago, maybe starting with TI's C30, I 
would assume that IIR filters are more popular than ever in terms of the 
number of applications they're used in.

That said, I would assume that most cell phones are implemented in 
fixed-point for reasons of both cost and power.  So, on a "units 
shipped" basis, FIR filters presumably still win.

Anyway, I guess it depends on what you mean by "popular", and even then, 
it's hard to know.  Assuming that people take whatever they read in an 
IEEE magazine as gospel, perhaps no such statement should be made at 
all.  Or, maybe something bland could be said like "the widespread 
availability of cheap floating-point processors has made IIR filters 
more popular than ever".  (If you use that golden prose, Rick, there's 
no need for attribution. ;-)

=g2
_____________________________________________________________________

Grant R. Griffin
Publisher of dspGuru                           http://www.dspguru.com
Iowegian International Corporation            http://www.iowegian.com
See http://www.iowegian.com/img/contact.gif for e-mail address

Grant Griffin <nospam@yahoo.com> writes:

> Now don't get me started on horrible zero-based assembly language, Randy...

I really want to start using "ith"-based addressing on my next assembly
language project. It would bring a whole new meaning to "orthogonal
instruction set." ...
-- 
%  Randy Yates                  % "Watching all the days go by...    
%% Fuquay-Varina, NC            %  Who are you and who am I?"
%%% 919-577-9882                % 'Mission (A World Record)', 
%%%% <yates@ieee.org>           % *A New World Record*, ELO
http://home.earthlink.net/~yatescr

Randy Yates wrote:
> R.Lyons@_BOGUS_ieee.org (Rick Lyons) writes:
> 
>> Hi Guys,
>>    I was asked to review a potential article 
>> for the IEEE Sig. Proc. magazine.  In that 
>> article the author implies that IIR (recursive) 
>> filters aren't as popular nowadays as they were in 
>> the past (say 10-20 years ago).  
>>
>> Now I'm no IIR filter designer, so I need your 
>> opinions.  Aren't IIR filters still as popular 
>> now for audio signal processing as they were 
>> 10 years ago?
> 
> I think the author(s) may be right. The situation is
> that we have much more powerful processors now than
> we did 10-20 years ago, so we can use brute-force 
> FIRs instead of IIRs. I don't think there's as big
> of a need for IIRs as there was in the past. Kinda
> like the way you don't see a lot of assembly language
> nowadays, either.

This argument ignores power consumption. A huge number of DSP 
applications are portable, and I don't just mean the obvious ones like 
cellphones. Even if you have lots of available MIPs, using them is bad 
in portable applications. Every instruction costs battery life. IIRs are 
energy efficient.

Steve

Rick Lyons skrev:
> Hi Guys,
>    I was asked to review a potential article
> for the IEEE Sig. Proc. magazine.  In that
> article the author implies that IIR (recursive)
> filters aren't as popular nowadays as they were in
> the past (say 10-20 years ago).
>
> Now I'm no IIR filter designer, so I need your
> opinions.  Aren't IIR filters still as popular
> now for audio signal processing as they were
> 10 years ago?

I don't know the relative popularity of IIRs and FIRs 10 years ago,
but I suspect IIRs will be a thing of the past in 10 year's time.

IIRs are difficult to design unless you have some dedicated
software. You don't have that unless you need it (like if you
are designing for low-power gadgets), so no one have that
software. In comparison, FIRs are easy to design using the
window function techniques. Eveybody can do that, you can
get there with a calculator an a little patience.

A few monts ago I implemented my own rudimentary IIR filter
design software. With one notable exception, the book by
Andreas Antoniou I have mentioned her a couple of times,
no texts on DSP do much more than mention IIRs in
passing. Some 80% of my job when implementing that
package was actually finding the relevant material (I only
found the Antoniou book after I was finished).

So yes, my impression is that the authors of your article
may be right, or at least there is a trend. IIR filters might
just be on their verge to become obsolete.

Rune

> the author implies that IIR (recursive)
> filters aren't as popular nowadays as they were in
> the past (say 10-20 years ago).

Probably goes back to the implication in 
old books that IIRs are better because they 
need less multiplications then FIR and 
since multiplications are timeconsuming
( shift & add ) so FIRs are no use.
"20-30 years ago" would be ok on that
line of argument.

MfG  JRD

Rick Lyons wrote:

> 
> Hi Guys,
>    I was asked to review a potential article
> for the IEEE Sig. Proc. magazine.  In that
> article the author implies that 
> IIR (recursive) filters aren't as popular 
> nowadays as they were in 
> the past (say 10-20 years ago).

if FIR filters have become more popular within these 10-20 years, and I'm
sure they have, then in consequence, IIR lost relative popularity.

> 
> Now I'm no IIR filter designer, so I need your
> opinions.  Aren't IIR filters still as popular
> now for audio signal processing as they were
> 10 years ago?

There are situations, where an IIR filter has advantages:
Design a butterworth HPF at 3Hz/10Hz (stop/pass) with sample rate of 48kS/s,
then try an equiripple FIR with similar quality.
IIR will require 5 biquads, while FIR is of order 18000... :-(

Depending on the filter requirements, you may or may not be able to replace
them by FIR filters. This will certainly change with increasing CPU speed
and memory space. 
However, IIRs will benefit from that, too, since the required precision
would be easier implemented.

I suspect, that IIRs won't disappear, and that in certain applications, FIR
filters wouldn't be able to replace them.

My opinion may be not audio-typical, since I'm working in the measurement
area. The above example seems to me as to be rather typical for audio
purposes, too.

Bernhard