|
 |
For a Printable PDF Version of this Newsletter
Click here to >> Download
Document << (approx. 168k)
|
Links to the latest
SeaSonde
software below:
Current Radial Software available: SSRadialSuiteCD_10R4SeaSonde10R4/
Current Combine Software available: SSCombineSuiteCD_10R3SeaSondeR3u3/
Note: You will be required to enter the
standard user name and password to enter
the customer section of SeaSonde.com
"The Sound of MUSIC" Heard
by SeaSondes
by Dr. Don Barrick
Everyone who has scratched beneath the surface of SeaSonde usage perhaps wondered
at the statement "SeaSonde bearings are determined using MUSIC". Or
maybe one hears about MUSIC parameters used in the Setup and Preferences files.
We explore what this is all about, and try to remove some of the "mystery
of MUSIC".
• History of MUSIC
Believe it or not, MUSIC (MUltiple SIgnal Classification) was devised by a very
bright Stanford researcher working for the CIA back during the Vietnam War [R.O.
Schmidt -- check our website to get his 1986 IEEE reprint]. They wanted to put
flush-mounted receive antennas on different parts of the skin of surveillance
aircraft. Receiving radio signals from enemy transmitters in the jungle below,
the goal was to find the direction of the source while flying by. This is called "direction
finding" (or DF). Conceptually, if one receives the same signal on antennas
at different locations, it ought to be possible to estimate its direction, just
as you or an animal turns its head and uses its brain to locate the direction
of a sound picked up by our ears. But what is the optimum algorithm to do this?
Others are available, but MUSIC is one of the best.
• How It Works in Two Paragraphs
Think of what is received on a given frequency as comprised of several possible
candidates: desired signal(s) and noise, the latter always being present. E.g.,
let there be two signals and noise present, heard on three antennas. Wouldn't
it be nice if one could devise some kind of "mathematical space" where
signals and noise were totally separated from each other? And where these clusters
of two signals and noise could be made to have identifiable properties when their
directions of arrival were properly chosen? Well, that's what MUSIC does.
The same signal heard on three antennas is
not totally different. When mixed up with
the other signals and noise, however, it is hard to separate them, much less
determine their directions. That's what
a matrix-based mathematical procedure
called "eigenfunction analysis" (a dreaded bane to graduate students
in the physical sciences) does for you. It creates three new "signals" --
called eigenvectors -- that are mutually perpendicular to each other, when you
get their directions correct. Each antenna has its own response vs. signal angle-of-arrival.
With SeaSondes we measure these with a transponder rather than naively relying
on textbook predictions which are rarely correct in the real world. Using these
antenna responses in matrix form, we step through direction-of-arrival angles
until the signal eigenvectors are perpendicular to the noise eigenvector. When
we have chosen the correct angles so that this happens, vector products become
zero, which is an easy algorithmic test to carry out.
• Why MUSIC and Direction Finding
with SeaSondes?
We at CODAR are the first to use MUSIC (or any direction finding) with radar.
Normally one forms and scans narrow beams to determine direction, like finding
an object in the dark with a flashlight. At HF, however, the size of antennas
required to do this (called phased arrays) is huge because of the long wavelength.
After a few early tests in the 60s, we decided this could not lead to a cost-effective
easy-to-use affordable system that people wanted. So we abandoned this (while
others did not), and CODAR was born. DF allows low-cost, inefficient tiny antennas
to be used in place of the massive antenna farms of yore, with no penalty in
performance.
Direction or bearing angle is determined for each current radial velocity that
is outputted from our Doppler analysis, because each velocity/Doppler bin is
an individual radio signal pulled out by the spectral processing. In our early
algorithms, we used a "least-squares" algorithm instead of MUSIC for
DF. But MUSIC had too many advantages to ignore. In fact, MUSIC DF was found
to be more accurate even for long phased array antennas than beam forming/scanning
[K. Laws, PhD Thesis, UCSC, 2001, available as PDF from our website]. Note that
MUSIC is not used in wave directional information processing with SeaSondes,
only for current mapping; relevant wave algorithms are described elsewhere.
|
 |
• Setup Choices, Decisions,
and
the Tradeoffs
Decisions must be made in MUSIC DF processing. The important one is: how many
directions are we actually receiving echoes from -- for the same Doppler (or
radial velocity) over a given circular range cell? Oceanographically, two is
a realistic upper limit, based on the resolution offered by HF radars in current
mapping. But the answer could be one. How to decide? If we guess wrong, the answer(s)
will be in error. That's where our "MUSIC parameters" come in. We start
off assuming both single and dual-angle situations, determine the respective
angles as described above, then run hypothesis tests to pick which situation
best fits the data: single or dual. You can read mathematical details of this
procedure in our patent [Barrick and Lipa, 1999, PDF on our website]; there is
no attempt to keep this hidden as in a "black box". To contrast two
examples that have been used, [10 5 8] are three parameters that favor single-angle
, while [80 40 1] are a set of parameters at the other extreme that favor the
dual-angle solution outcome.
Suppose you push toward single angle when the actual current pattern is dual-angle?
Then you will get angle gaps in your radial map coverage. That's not good. Gaps
are especially aggravated if the measured antenna patterns are very distorted
from ideal, or if the receive channel amplitude or phase balances drift over
time and are not accounted for. On the other hand, if dual-angle is obtained
when the situation demands single-angle, these wrong answers will bias the mapped
results.
• How Did We Arrive at Default Values -
When Might One Change Them?
Two methods were used over the years: (i) Extensive simulations; (ii) Comparisons
of SeaSonde outputs with independent measurements like ADCPs. With simulations,
you know the answer going in, and you devise input flow patterns to resemble
realistic circulation features commonly seen. But there's nothing like independent
comparisons in the real world. Many studies of both simulations as well as comparisons
have been done by our customers, lending greater credibility to our recommended
MUSIC parameters. It's always good to have outside studies like this rather than
trust the manufacturer for everything, isn't it? After all, scientific claims
should be reproducible by others if they are to be credible.
Right now, based on our experience, we recommend as default MUSIC parameters
[40 20 2]. If you want to study differences between two parameter sets on you
own data, you can re-process offline both ways. Contact us so we can advise how
to set this up and what parameters to try. And let us know if you want to learn
more of what others have done on this subject.
|
|
 |
Tech's Corner:
Protecting Your SeaSondes® With a "Smart" UPS
Configuration
by Chad Whelan, Sr. Field Engineer
What is a UPS and why do I need one?
As all of us who work with high-tech equipment
know, a stable supply of power is of paramount
importance to reliable, uninterrupted data collection. UPS is
an acronym for
Uninterruptible Power Supply. By using a UPS,
you give your equipment a reserve of power from a backup source
that filters the incoming
power and allows for controlled
shutdowns in the event of an extended outage.
The amount of time a piece of equipment can
run on a UPS during a power outage depends on the size of the
UPS battery and the
power consumption or draw of the equipment.
The Transmitter and Receiver chassis have somewhat different
requirements of a UPS
than the computer.
Transmitter and Receiver
While they have been designed to withstand
some power line fluctuation, both the transmitter and the receiver
are sophisticated pieces of equipment and should be afforded
additional protection. There are a number of power-related circumstances
that
can cause damage, wear or undesirable performance.
Brownouts, power surges, voltage spikes and
short, momentary power outages are the most
common types of damaging power issues.
Since the transmitter is a constant power device,
if the voltage drops to a low, non-zero value
(a brownout) or it ramps down
to zero volts slowly, the transmitter will
try and compensate by drawing more current.
This can blow out the fuse that protects
its internal components. Unfortunately, when
this happens, someone needs to visit the site
and replace the fuse manually. Frequent
momentary losses of power are another concern
as they can cause the internal modules to restart
repeatedly and behave in unexpected
ways. Things like USB communications between
computer and receiver have been known to be
affected by frequent module restarts. Power
surges and spikes, if strong enough, can also
cause data loss if they blow fuses or damage
components.
Computer
The computer is the interface between the user
and the electronics and in addition to collecting
and processing data, it can be used for diagnostics
and monitoring performance.
During a long power outage, it may be desirable
to keep the computer active even if the transmitter
and receiver have drained their
UPS. It provides the user much
more information to be able to log on and see
that the UPS is not receiving power rather
than not being able to access the computer
at all. Computers also do not typically
respond well to losing power frequently, sometimes
even failing to reboot when power is restored.
It has been reported that in
some cases, cutting power to the computer has
caused the clock to reset and lose the current
time. A controlled shutdown is
always a better solution than just cutting
power. By monitoring the UPS storage, a computer
can shut itself down to protect itself,
but also to conserve power and set a
reboot time.
|
Proposed Setup
Because of their different needs, one possible solution is to use one UPS for
the SeaSonde® electronics and a separate one for the computer.
The UPS for the electronics should be of at least 1 kVA but preferably 1.5 -
2 kVA and does not need to have a USB interface. With AC power, 1.25 kVA is approximately
1 kW. UPS’s are typically rated in kVA. The UPS for the computer does not
need to have as high capacity, but should have enough storage to last the computer
for several hours. It should also have a USB interface that is compatible with
OS X, like the APC SmartUPS® line of products. This configuration will provide
the electronics with stable power through outages and power line fluctuations
as well as a clean voltage drop to zero volts when the UPS storage has been spent.
It will also provide the computer with stable power and controlled shut-downs
before power is removed. We have developed a script that will monitor the UPS
and shutdown the computer periodically to conserve power as well as prevent an
immediate power loss while the OS and software are running. When AC power is
gone and the UPS storage reaches a user-defined power reserve, the script will
schedule a startup at a user-defined time or interval and then shutdown properly.
When the system starts up again, it will determine whether AC power has returned.
If power has not returned, the computer will remain on for a user-defined interval
and then go through the process again. It will continue this way until AC power
has returned or the UPS has reached another user-defined threshold reserve, below
which it will remain off until power is restored. If a communication device,
like a router, is also on UPS power, it will allow the user to log in during
one of these reboots to check on the system.
Please contact CODAR Ocean Sensors
Support  to
find out more about this or other UPS options. |
|
More Tech's Corner:
Important Message for GPIB equipped SeaSonde® users! USB Upgrades Available for GPIB
Systems!
CODAR Ocean Sensors (COS) will be phasing out
all support for SeaSondes using the obsolete
GPIB (General Purpose Interface Bus) over the next six months.
The good news is that
COS will be offering specially priced USB upgrades
to encourage the few remaining GPIB system owners to make the
leap to the
USB interface which is now an industry standard.
Two specially priced upgrade packages will
be available through June 30, 2006. One is a basic upgrade to
the USB bus. This will allow older SeaSondes to take advantage
of all of our new OS X based software. A second upgrade option
includes the addition of a GPS timing module and built-in wattmeter.
The GPS timing module allows multiple SeaSondes to operate simultaneously
while sharing the same frequency. This upgrade option also provides
remote monitoring of output power.
If you are one of the half-dozen or so remaining "loyal" GPIB
system owners and are interested in adding
years more life to your system along with a one year warranty
renewal, please contact
for pricing information.
Help is Available
if You Ask
CODAR has a dedicated and experienced Support
team available to help customers with potential problems, and
can provide instant feedback if you suspect something is not
right with your data.
Examples of situations that might precipitate
contacting our support staff: (1) Significant variations in maximum
coverage, e.g., radial map patterns fluctuate by 40% over a 24
hour period. (2) Noticeable, regular gap regions appear in specific
positions on your maps. (3) Wild vectors occasionally are spotted
in circular bands at ranges from one or the other site from 100-130
km (for Long-Range systems).
Any change from data outputs that you had been
getting, or from what you expected, can probably be remedied
... But we need to know what you are seeing that you don't like!
We will then swing into action, diagnose your problem, fix it,
and educate you as to what happened.
At the lower operating frequency bands, late-afternoon
and nighttime radio interference is known to occur, even in systems
that had not been seeing this before. Nighttime is when radio
broadcasters use the lower HF band, and usually for only a few
hours. Such interference appears as noise to our processor. Maximum
range will decrease during these periods, even though vectors
closer in are perfectly valid. Remedies are to shift positions
within your present authorized frequency band, or move to another
frequency that you were granted. We can diagnose this immediately,
and teach you how to do it also.
PLEASE NOTIFY US OF ANY PLANNED CHANGES BEFORE
OR WHEN YOU
MAKE THEM
|
Often a customer will change frequency bands,
move one of the antennas, or rotate the receive antenna, perhaps
for a very good reason.
We can often advise you before such a
change what kind of "domino effect" this could result
in. Also, immediately after the change, we
can monitor your system to verify that it is operating correctly,
or make recommendations
for additional modifications. Again, we will
keep you informed as to what we find. Nearly 85% of problems
happen as a result
of a change to system settings (software
or hardware) that we are unaware of. We only want your systems
to be producing
the best possible data for you. When that
happens, we both look good. But we can only do that with input
from you.
Don't hold back!
Are you up to date?
SeaSonde10 users should be running:
SeaSondeRadialSuite10 Release 3 with
Updater3 installed
SeaSondeCombineSuite10 Release 3 with Updater3 installed
Mac OSX 10.3.9
Timbuktu 7.0.4 (required for latest Timbuktu
scripting)
OS9 SeaSonde users should be running 4.4f6
Mac OS 9.2.2
If you have any questions,
please email us |
1914 Plymouth Street
Mountain View, CA 94043 USA
Phone: +1 (408) 773-8240
Fax: +1 (408) 773-0514
www.codaros.com |
|
|
|
