research problem

Hullo, it's only me...

This is a bit lazy on my part, but I am rooting around for a few
suggestions for some hobbyist research. When I joined my current job, I
didn't read the contract carefully, and aparently any research on
digital communications I do in my spare time belongs to the company (is
this even legal!?). I want to keep up research and publication, because
eventually I would like to go back into academics, so if I cant work on
digital communications in my spare time, would any of ye recommend
research topics which I might enjoy, which would require a similiar set
of skills for that used in digital comms?

I have been thinking along the lines of the common threads on this
group, which include speech and audio processing, image and video
processing, control theory and applications, and occasionally
biomedical signal processing. However, I am worried about some of the
obstacles to entry into some of these fields. For example I believe
speech and audio processing are very mature fields (although I still
havent seen any good voice recognition software :-). Would i have more
luck in producing worthwhile research if I picked a field which may
benefit more from the insights of a communications engineer? (I have
some crazy ideas about applying the tools of comms theory to the
financial markets, but something tells me I missed that bandwagon long
ago :-).

Of course, I will only find what I am looking for if I get off my arse
and do some propper research, but it seems that there are so many
topics out there, that I could go down a lot of dead ends... feels like
I am chosing my PhD topic all over again :-). I think I would just like
to add another string to my bow that would complement my work in
digital comms, and if you have any suggestions they would be
appreciated... what are the hot research topics in image and video
right now anyway???

(just so I dont seem to be taking without giving, for anyone who is
interested the hot topics in digital comms right now are MIMO comms,
cross-layer (MAC/PHY) design, ultra wideband, multi-carrier comms...
topics like wavelets seem to be going out of fashion, they were all the
rage for about a decade, you dont hear much about precoding these days
either. If I was starting research in digital comms now, I would be
looking at network information theory, multi-user, multiple access
techniques, space-time modulation and coding and the others mentioned
above. Of course there is always some nut working on what everyone
thinks is left-field, but who turns out to be onto a winner... like
those guys from Brest who invented Turbo coding at the beginning of the
nineties.... I'd love to know what the 2005 Eureka moment will be but
if I knew that, I'd quit my job and go work on it :-)

Jeepers, this was ment to be a short post....
Slainte
Porterboy

porterboy76@yahoo.com skrev:
> Hullo, it's only me...
>
> This is a bit lazy on my part, but I am rooting around for a few
> suggestions for some hobbyist research. When I joined my current job, I
> didn't read the contract carefully, and aparently any research on
> digital communications I do in my spare time belongs to the company (is
> this even legal!?).

It is legal. Many carreers have been destroyed because of people
not reading and understanding that part of the contract.

> I want to keep up research and publication, because
> eventually I would like to go back into academics, so if I cant work on
> digital communications in my spare time, would any of ye recommend
> research topics which I might enjoy, which would require a similiar set
> of skills for that used in digital comms?

Oh, y'know, the usual: Maths, electronics, systems,... but
you knew that.

> I have been thinking along the lines of the common threads on this
> group, which include speech and audio processing, image and video
> processing, control theory and applications, and occasionally
> biomedical signal processing. However, I am worried about some of the
> obstacles to entry into some of these fields. For example I believe
> speech and audio processing are very mature fields (although I still
> havent seen any good voice recognition software :-). Would i have more
> luck in producing worthwhile research if I picked a field which may
> benefit more from the insights of a communications engineer? (I have
> some crazy ideas about applying the tools of comms theory to the
> financial markets, but something tells me I missed that bandwagon long
> ago :-).

The one good part about communication technology, is that the
results are measurable. You test a new idea, and either it works
or it doesn't. Either the benefit ouutweighs the cost of
implementation, or they don't.

Some of the other research fields you mention, are very vague on
such stuff, so it might be an advantage for you to read up
on such fields to compare to your own, and see the difference
from communications.

One difficult but mature -- and results-oriented -- fioeld is
seismic exploration. It takes a bit of physics to get in there,
but it depends heavily on signal processing and the analysis
works in a completely different way.

> Of course, I will only find what I am looking for if I get off my arse
> and do some propper research, but it seems that there are so many
> topics out there, that I could go down a lot of dead ends... feels like
> I am chosing my PhD topic all over again :-). I think I would just like
> to add another string to my bow that would complement my work in
> digital comms, and if you have any suggestions they would be
> appreciated... what are the hot research topics in image and video
> right now anyway???

Do you have access to matlab? Work your way through McClellan,
Oppenheim, Schafer & al's collection of matlab projects in
signal processing.

> (just so I dont seem to be taking without giving, for anyone who is
> interested the hot topics in digital comms right now are MIMO comms,
> cross-layer (MAC/PHY) design, ultra wideband, multi-carrier comms...
> topics like wavelets seem to be going out of fashion, they were all the
> rage for about a decade, you dont hear much about precoding these days
> either. If I was starting research in digital comms now, I would be
> looking at network information theory, multi-user, multiple access
> techniques, space-time modulation and coding and the others mentioned
> above. Of course there is always some nut working on what everyone
> thinks is left-field, but who turns out to be onto a winner... like
> those guys from Brest who invented Turbo coding at the beginning of the
> nineties.... I'd love to know what the 2005 Eureka moment will be but
> if I knew that, I'd quit my job and go work on it :-)

Well, I'd suggest you stay well away from anything that has to
do with subspace-based spatial processing of antennas (methods
like ESPRIT and MUSIC applied to high resolution spatial
beamforming). Those kinds of things are dead ducks, the methods
rely on more data than signal sources, i.e. there must be at
least as many single elements in the antenna ast ther will ever
be users. If there are more users than sensor elements, the whole
system will break down for all users. Not something one would
like to introduce in dynamic com systems that are expected to
work reliably.

Very basic stuff, really, but there used to be lots of papers
published on those sorts of techniques in those sorts of
applications.

> Jeepers, this was ment to be a short post....
> Slainte
> Porterboy

Rune

Rune Allnor wrote:
> 
> porterboy76@yahoo.com skrev:
> > Hullo, it's only me...
> >
> > This is a bit lazy on my part, but I am rooting around for a few
> > suggestions for some hobbyist research. When I joined my current job, I
> > didn't read the contract carefully, and aparently any research on
> > digital communications I do in my spare time belongs to the company (is
> > this even legal!?).
> 
> It is legal. Many carreers have been destroyed because of people
> not reading and understanding that part of the contract.


It depends on jurisdiction. I'm almost certain that is not legal here
in Australia. I've also been very careful to make sure my ability
to do outside projects are explicity allowed.

Erik
-- 
+-----------------------------------------------------------+
  Erik de Castro Lopo  nospam@mega-nerd.com (Yes it's valid)
+-----------------------------------------------------------+
"The music business is a cruel and shallow money trench, a long
 plastic hallway where thieves and pimps run free, and good men
 die like dogs. There's also a negative side."
   -- Hunter S. Thompson

Erik de Castro Lopo <nospam@mega-nerd.com> writes:

> It depends on jurisdiction. I'm almost certain that is not legal here
> in Australia. I've also been very careful to make sure my ability
> to do outside projects are explicity allowed.

I am certain that it is legal in Australia, at least with respect to
the intellectual property generated (copyright, patents etc).  As an
engineer, virtually all intellectual property that you generate in the
employer's area of expertise is owned by your employer.

http://www.brainbox.com.au/members/brainbox/home.nsf/0/513652F84BA8FDC349256D31002CFBCA?opendocument

If your outside-work projects generate revenue from copyright or
patents, then your employer pretty much owns the income unless you
have a specific agreement.

If the projects generate income by provision of services, then it's
not an issue --- you own the income and your customer owns the
results.

Ciao,

Peter K.

PS: IANAL!!!!! :-)

<porterboy76@yahoo.com> wrote in message
news:1124487352.366989.59800@g49g2000cwa.googlegroups.com...
> Hullo, it's only me...
>
> This is a bit lazy on my part, but I am rooting around for a few
> suggestions for some hobbyist research. When I joined my current job, I
> didn't read the contract carefully, and aparently any research on
> digital communications I do in my spare time belongs to the company (is
> this even legal!?). I want to keep up research and publication, because
> eventually I would like to go back into academics, so if I cant work on
> digital communications in my spare time, would any of ye recommend
> research topics which I might enjoy, which would require a similiar set
> of skills for that used in digital comms?
>
> I have been thinking along the lines of the common threads on this
> group, which include speech and audio processing, image and video
> processing, control theory and applications, and occasionally
> biomedical signal processing. However, I am worried about some of the
> obstacles to entry into some of these fields. For example I believe
> speech and audio processing are very mature fields (although I still
> havent seen any good voice recognition software :-). Would i have more
> luck in producing worthwhile research if I picked a field which may
> benefit more from the insights of a communications engineer? (I have
> some crazy ideas about applying the tools of comms theory to the
> financial markets, but something tells me I missed that bandwagon long
> ago :-).
>
> Of course, I will only find what I am looking for if I get off my arse
> and do some propper research, but it seems that there are so many
> topics out there, that I could go down a lot of dead ends... feels like
> I am chosing my PhD topic all over again :-). I think I would just like
> to add another string to my bow that would complement my work in
> digital comms, and if you have any suggestions they would be
> appreciated... what are the hot research topics in image and video
> right now anyway???
>
> (just so I dont seem to be taking without giving, for anyone who is
> interested the hot topics in digital comms right now are MIMO comms,
> cross-layer (MAC/PHY) design, ultra wideband, multi-carrier comms...
> topics like wavelets seem to be going out of fashion, they were all the
> rage for about a decade, you dont hear much about precoding these days
> either. If I was starting research in digital comms now, I would be
> looking at network information theory, multi-user, multiple access
> techniques, space-time modulation and coding and the others mentioned
> above. Of course there is always some nut working on what everyone
> thinks is left-field, but who turns out to be onto a winner... like
> those guys from Brest who invented Turbo coding at the beginning of the
> nineties.... I'd love to know what the 2005 Eureka moment will be but
> if I knew that, I'd quit my job and go work on it :-)
>
> Jeepers, this was ment to be a short post....
> Slainte
> Porterboy
>

Speech recognition - with good training Dragon Naturally Speaking is nearly
100% accurate.
Even the Microsoft engine is quite good for limied vocab - speak
independent.

Shytot

>what are the hot research topics in image and video
>right now anyway???

I personally think the best way to gather the general trend around cosumer
video(which is the largest chunk of the video market)is to go to your
nearest Authorised Sony and Samsung showrooms and ask them to explain the
technology behind the latest flat screen HDTV's. Sony has about four
competing technologies right now. And i remember hearing about a
technology called DCR ,which ended up being some combination of comb
filters! And if am not mistaken anything related to surveilance like
"motion estimation and tracking" are buzz words too. I heard of a cool
demo where this pc based logitech web-cam, once focussed on you the user,
will follow you in all directions and axes as you move your head around.
Pretty cool, eh? Identify terrorists and keep em locked!

--Bhooshan
		
This message was sent using the Comp.DSP web interface on
www.DSPRelated.com

Bhooshan Iyer skrev:
> Pretty cool, eh? Identify terrorists and keep em locked!

How do you "Identify terrorists" by means of DSP?

Rune

"Rune Allnor" <allnor@tele.ntnu.no> wrote in message 
news:1124571135.451148.178680@g49g2000cwa.googlegroups.com...
>
> Bhooshan Iyer skrev:
>> Pretty cool, eh? Identify terrorists and keep em locked!
>
> How do you "Identify terrorists" by means of DSP?
>
> Rune
>

I guess you do it by identifying the "signals" they give off.

Clay

Clay S. Turner skrev:
> "Rune Allnor" <allnor@tele.ntnu.no> wrote in message
> news:1124571135.451148.178680@g49g2000cwa.googlegroups.com...
> >
> > Bhooshan Iyer skrev:
> >> Pretty cool, eh? Identify terrorists and keep em locked!
> >
> > How do you "Identify terrorists" by means of DSP?
> >
> > Rune
> >
>
> I guess you do it by identifying the "signals" they give off.
>
> Clay

You'd better make sure you get your algorithm properly
debugged before testing it:

http://news.bbc.co.uk/1/hi/uk/4711021.stm

Rune

Rune Allnor wrote:
> Bhooshan Iyer skrev:
> 
>>Pretty cool, eh? Identify terrorists and keep em locked!
> 
> 
> How do you "Identify terrorists" by means of DSP?

Shifty eyes.

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������