Last update 07/07/2009
( links to DARPA photos and videos )
2004 DARPA Grand Challenge Communications For the
Event
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The equipment and techniques used by
The Defense Advanced Projects Agency is an agency
of the United States Department of Defense, and is charged with identifying new
technologies useful to the Armed Forces of The United States and initially developing
those technologies.
Congress has mandated that the mundane overland
transportation tasks of the military, such as supply convoys; shall be accomplished by
autonomous vehicles before 2020. In order for these vehicle systems to be effective, they
must be able to navigate and sense obstacles in their path without human intervention. In
other words, once sent on a mission they are on their own. DARPA issued a challenge to non defense industry
entities and individuals to develop and showcase these guidance technologies: "The
challenge is to build a robotic vehicle that can successfully navigate a 150 mile course
across the A wide variety of entries answered the Challenge, ranging from teams sponsored by major universities to garage robot enthusiasts. Of over 100 entries, 25 were chosen to compete. Of those 25 accepted, 13 Bots satisfactorily completed a comprehensive QID (Qualification, Inspection, and Demonstration) event. The prospect of a group of robot vehicles
loose in the desert raises many concerns, such as safety, traffic and course
control, environmental and ecological monitoring, and general race operations. Several hundred people were involved along the
course for road closings and safety, as well as 25 biologists charged with tracking and
wrangling several endangered species such as the Desert Tortoise and The Wild Burro. A large number of Command and Control personnel
needed to access the radio system in the confined space of the Command and Multiple repeater locations would be required, and
Land Mobile Radio voice communications, Vehicle tracking data, computer networking, and
telephone circuits had to be distributed along the 150 mile race course across the A VHF Aeronautical radio interface able to
communicate with Air Assets (helicopters) over the entire race course would be required. For this discussion, the system will be broken down into subsystems:
Command and Control (Dispatch) Each Bot would have a chase vehicle
(CV) following it in close physical proximity. Each CV was required to be in constant
radio contact with a corresponding Judge at the Command and Operations Center (COC) Finish
Line location in It would be necessary to provide each Judge with
the ability to access multiple channels in whatever configuration the Judge deemed
necessary, and permit easy channel switching and reconfiguration of those channels. The
field of entries had been limited by DARPA to 25-30 vehicles. So, 25-30 hard comm
positions for Judges would be provided at the Primm COC. A Race Director would oversee Operations at the
Primm COC. The Director would be moving around the COC overseeing the Judges, and would
have one of the few radios in the room on headset. Five comm positions at the Stoddard SAF would be
required, and comm positions for Environmental monitoring and interface to Local, County,
and State Agencies from both The prospect of bridging 30 or more Tone Remote
consoles together at the COC was daunting, considering the audio level jumps that would
occur as users switched on and off the various channel busses, not to mention the table
real estate that would be required for them. Then, bridging additional consoles from a
location at the other end of the course was an additional level of unwanted complexity. However, since there would be distributed
transmitters along the course, Tone remote control (TRC) would be used to control them so
TRC capability would still be needed.
The C6200 Console product from Telex Vega is a full-featured multi-channel dispatch console which is fully implemented with Digital Signal Processing (DSP). The native interconnect format for the C6200 is by way of Ethernet and Voice-Over-Internet-Protocol (VoIP). Analog Line Interface modules provide interconnection to analog radio systems.
Telex Vega also offers a console emulation
software application called C-Soft
that runs on a PC. The PC interface can be custom configured to the requirements of the
user, and the virtual control surface can be customized so that only the required controls
and channels appear to a given user. With the Telex Vega System, an unlimited number of
consoles and console emulation PCs on a common network can operate as an integrated
system. Any PC on the network running Windows XP or Windows 2000 and containing a sound
card and headset can be used as a full-function dispatch console, or as a monitor point.
In fact, the VoIP traffic can even be packaged and routed over the Internet, and with
sufficient connection bandwidth the radio system traffic can be monitored at any point on
the globe. The Telex/Vega system was a perfect fit for this
project. Each Judge at the COC would have a PC to run the
vehicle tracking software and would already be networked. The Judge could have the
tracking software in the foreground for best visibility and the C-Soft application in the
background. If channel changes or volume changes were required, the C-Soft window could be
brought to the foreground, the changes made, and then returned to the background. C-Soft
provides an interface via the serial port on the PC for an external PTT switch, which
allows PTT operation of the application even if it is not in the foreground. Simple PC-type headsets eliminated the din of
radio traffic in the COC. The left ear heard the select audio from the primary channel,
and the right ear heard the unselect audio from the other channels. The individual user
could mix his or her own ratios of unselected channels, and independently adjust the
volume of select (long horizontal slider) and unselect audio (short horizontal slider). Actual Event C-Soft Panel over Map of Microwave
Paths In the COC, workstations were arranged around the perimeter of the area and in the center of the area. The Primm COC in a quiet moment
In the initial planning stages of the
communications system, a mixture of common-carrier fiberoptic and microwave T1 circuits
was envisioned to interconnect the remote transmitter and receiver sites, with private
short-haul microwave circuits to locations where common-carrier circuits could not be
obtained. As the costing came in for the common-carrier
circuits, it became clear that it would be more cost-effective to do a complete private
microwave network. At a late date, DARPA also decided that the vehicle tracking system would have to become near-real time. The enhancements to the tracking system would require significant bandwidth downrange, and would increase common-carrier costs by an order of magnitude. Several methodologies would be used by network contractor CenGen to bring multiple data streams from downrange into Primm. A robust and reliable microwave system would be
required which would be capable of handling a mixture of traffic, including channelized
T1, LAN, and routed RS232 data.
Comprehensive planning discussions with ADTRANs excellent applications engineering
department were essential to the successful deployment of this network. Since the microwave network would be temporary,
there were limits to the size of the antennas that could be deployed for the network due
to the requirement to use our temporary non-penetrating roof mountings (we call them
"Lunar Rovers"). However, the hops between the radio system sites were as long
as 42 miles. 4 foot Andrew
P4-57W antennas were chosen, and modified by In transit, four of them had bounced out of their frames and had pancaked on top of each other in the truck. A couple of careful taps and the the dings came out and the path worked perfectly.
Stoddard SAF
Primm Comm, NOC, COC (L-R) The microwave sites, which were also UHF radio
sites, were arranged as T-1 repeaters with TRACER radios connected back-to-back. At each site, the T-1s were cabled down to temporary equipment shelters which contained the UHF radio equipment. In each shelter, ADTRAN TSU600 T-1 channel banks with drop and insert capability were used to drop the E&M circuits for the site and pass on the E&M circuits for sites downrange. Due to the excellent processing power of the ADTRAN equipment, no perceptible latency was introduced into the audio between the sites, even after passing through multiple channel banks and multiple drop and insert steps. Temporary Shelter and Roof Mounted Microwave and Radio Antennas Site
Supervision, Backup Power, and Facility Power CenGen provided a Cisco router at each site which
connected to one of the T-1s for tracking data and networking. CenGen also provided
an APC
1.4 KVA UPS at each site to hold up the router, the tracking data equipment,
the radio channel bank, and the microwave equipment. Tracking data from the vehicle
tracking system provided by Omnitech Robotics
was interfaced to the microwave network and routed up and downrange. Site condition
telemetry and path health was derived from router link up/down status and packet errors,
and the UPS status via Ethernet indicated utility power loss or presence. The Motorola repeaters were backed up with
individual battery banks per repeater. Each remote site system was then further backed up
with a portable diesel generator provided by Aggreko.
If power was lost, the battery backup systems would hold up the site until a technician
stationed at the site for the event could get the generator started and change over to it. A 150KW auto-start auto-transfer genset provided
by Aggreko at the Primm COC provided backup power for the Command, Network, and Comm
facilities. Enough fuel was provided for all of the generators
so that 24 hours of operation with all systems running could be achieved if all utility
power between UHF Repeater Network and The Mohave Desert Terrain DARPA wanted the most comprehensive coverage of
the course which was practical. It went something like this: DARPA: Whats the difference in
cost? DARPA: anywhere is
adequate. Coverage was defined by DARPA had determined that 4 course-wide nets would
be required, and that users would need to access any of the four nets at any point on the
course. The terrain in the Mohave is very rugged with elevation variations along the route
of from 800 ASL to 5000 ASL. Inquiries with public safety agencies with these
areas in their jurisdictions indicated that radio coverage would be challenging. American Tower Corp. was of significant assistance in this regard, and seven ATC sites were selected. We really cannot say enough about the cooperation and help we got from Sam Drogin and Ron Mihares of ATC. The site selections were made based on proximity to the course, the probable route of which was known by November of 2003. Subsequent evaluations and efforts to mitigate costs reduced the number of ATC sites needed to five. Of those, four would be UHF repeater and microwave sites, and one would be a microwave repeater site only. RF planning for the 5 sites began in November with
path modeling for the microwave and coverage modeling for the UHF channels. When the final
selection of the sites was completed, Doug Thompson of Cara Enterprises handled filing the
complex STA for the UHF channels. Because of the proximity of the sites to each other (ranging from 22 - 42 miles apart) and the elevations of the sites (as much as 6000 ft ASL), it was certain that if all the radio sites were on identical repeater talk-out frequencies there would be significant co-channel interference. In fact, from a couple of the sites, ALL the sites could be heard. Route Map of the Microwave System with UHF Radio Sites However, the proximity of the sites was ideal for implementation of a multi-channel, multi-site voting system. A channel scheme was developed which allowed all the sites to receive on the same set of four receive frequencies, but allocated transmit frequencies in even/odd sets at alternating sites. This allowed 80 miles between sites that used identical transmit frequencies. The harsh terrain and mountains of the Mohave would provide sufficient isolation between sites using the same transmit channels in all but a few locations. Actual Elevation Profile of Repeater Sites Across
the Mohave The Motorola Digitac comparators have extensive
audio input and output flexibility and readily accommodated the complex audio paths
required to interconnect the system and the dispatch console chain; and can accommodate 8
receive site inputs per system channel. The Digitacs are part of an existing system owned
by Four course-wide channels were required to support
event operations over the entire length of the course, and six simplex local channels
would be used at areas where concentrated communications were to take place, such as road
closings and crossings. Of those six local channels, five were allocated for specific
purposes, and the sixth was allocated for use either as an independent channel or it could
be used to relay through vehicular repeaters in 3rd St. R & D vehicles. Build-Out, Installation and Turn-on Build-out of the system began in November of 2003
and was largely completed in time to load and leave for the desert on The site equipment was contained in three portable
communications shelters and one 4WD truck and all traveled on a flat-bed semi-trailer. Shelters and Truck Site Loaded for the return trip
.Testing and Tweaking During February, several test excercises were run, and the system was tweaked and fine-tuned for best performance. Interesting and unanticipated effects were noted after tweaking was complete. We knew that the voting system would be an important element to the system, but we really thought that it would merely be a means to utilize receivers at a number of locations. We thought that if any mobile was within 15 or 20 miles of a receiver that it would be sensed by that nearest receiver and that would be that. Due to the topography of the Mohave and the shadows of the mountains, and the fact that the UHF signals would duct long distances through the mountains; there were times when the active receiver would be at a great distance from the transmitting mobile. There were many locations where a nearby site would be totally shadowed, but a site 60 or 80 miles away had the mobile full-quieting. The voting system was essential for successful operation. We had mostly used the voting system with portables in the past, not so much with mobiles. The additional power (110 watts vs. 4 watts) meant that the mobiles were able to activate very distant receivers for clear coverage where portables might not have. We also became aware that the portables have very different audio quality than the mobiles. Motorola portables typically have very smooth audio response. The Spectra, GM300, and Maxtrac mobiles on the other hand, have audio response tailoring in the circuits of their palm microphones which give them some additional boost in the upper end of the voice range for enhanced intelligibility. That boost in the high frequencies caused the comparators to evaluate the audio as being noisy under some full-quieting conditions especially with some user voice characteristics. We developed an improved method of setting reference levels to eliminate those effects. Interestingly, mobiles which were equipped with David Clark Headsets, which have VERY smooth microphone elements without additional response tailoring, did not cause these effects even with "hostile" voice characteristics.
Land Mobile Radio Inventory Allocations For the Control vehicles (CVs), 25 Standard CVs and 5 Super
CVs were equipped with this unit complement. Standard CVs were each assigned a specific
Bot to control. A 900 MHz RF E-Stop remote control link provided by Omnitech
Robotics connected each CV to its respective Bot. If the Bot left the course
or appeared to be exhibiting dangerous behavior, the CV could E-Stop the Bot either
in a pause mode or in a shut-down-and-disable mode. The Super CVs had master control E-Stop systems in them which
could take over for any CV which experienced mechanical difficulties. The Super CVs
were stationed along the course so that they could quickly enter the course, catch up with
the Bot and CV and take over for the compromised CV. The Spectra was used as the primary course-wide
radio on the primary nets. The alpha-numeric display was essential in the bright sunlight
of the vehicle cab and was readable by the radio operator even under the rough ride
conditions of the course. Spectral purity of the mobile radios, especially
the high-power Spectras, was essential since the first harmonic of the UHF frequencies
could have had a very deleterious effect on the critical 900 MHz E-Stop systems. The
antenna systems for the UHF radios and the 900 MHz radios were only about 18 apart
on the vehicle. No ill effects were observed in testing and in use. The GM300 radio was assigned primarily to monitor
the course-wide Safety channel, and to provide a backup in the event that a Spectra or a
portable had a failure during the event. The portable was used for local comm between
nearby Control Vehicles to coordinate passing, and for local communications with road
crossing crews as the vehicles approached them as well as with observation helicopters
overhead. The right-seat Judge controlled the local portable
and the Safety Net radio, and the backseater operated the Spectra for contact with the COC
Judges and Race Director. Portable
(Handheld) Unit Allocations Along the course, 55 portables were allocated to
road crossings and at locations where the Bots were to travel public roads. 20
Monitor Point observers watched critical locations where the Bots would come into
close physical proximity to power lines or other utility infrastructure. 10 Master
Transmitter operators could take over control of a Bot from a CV if dust or distance
obscured the CVs view. Race officials at the Start and Finish lines, COC
and SAF were issued 30 portables and headsets for local coordination of those activities. Five Air Asset helicopters, in addition to their
aeronautical radios, were issued Motorola Portables with SetCom double-muff high noise
level Series 7 headsets so that they could communicate with the CVs below them. An aeronautical radio with tone remote interface
was placed at the mid-course radio site at an elevation of 4000 AGL. That
Aeronautical radio could be accessed by any Comm position to directly communicate with the
helicopters from either the COC in Primm or the SAF in Stoddard. The biology staff, charged with monitoring and
protecting the Desert Tortoises, was issued 25 portables. The Tortoises exhibited an
uncanny knack for being discovered in the few locations along the course where portable
coverage was below specification, so key biology lead personnel were issued GM300 mobiles
for their vehicles. All told, 246 portable radios were issued to users along the 150 mile course, along with 37 GM300 mobiles and 35 Spectra Mobiles. All the equipment performed beyond expectations and the communications system and ancillary systems were judged to be rock solid by DARPA personnel. The Human
Interface Minimal training time was available for all the
various people who would use the system. Because of CenGens excellent work, computer
users were basically unaware of the length and breadth of the network. The communicators, both mobile and in the COC/SAF
had widely varying levels of experience with radio systems and operation, ranging from
none to significant experience with tactical battlefield communications. Training for the
communicators (judges) who were to use the C-Soft application was very time-limited, and
except for two short sessions to acquaint the users with the basic on-screen controls,
their only real training was hands-on during two practice runs. Because the C-Soft screen can be designed for the
specific application, non-essential controls were deleted from the screen interface and it
was reduced to the bare minimum. The users were able to quickly learn and adapt to the
interface and use it effectively. The behavior of a voting system took a little time
to get used to, as the mobile users had to be made aware of the need to push the
button, take a breath, then talk to give the system time to make a receiver
decision; and to give the transmitter chain time to come up on a console push to talk. It was clear that most users were conditioned to
military point-to-point simplex PTT behavior of walkies in close proximity, rather than
the setup times involved in repeater systems with tone squelch and keying tone sequences. However, the users quickly adapted to the need for
good technique and it ceased to be a problem. Keyup and vote setup was still much quicker
than a trunked system. As previously mentioned, it was necessary to
reduce the number of radios in the COC and SAF locations to either none, or to only units
with headsets due to the audio latency introduced by the VoIP systems. It is in the nature
of VoIP for there to be some minor audio latency, generally on the order of 300ms or so.
This latency was not a factor to any user not actually in the COC or SAF, but could be a
source of confusion to a radio user in the COC, or to a communicator who was hearing his
voice come out of a radio 300ms late. Since the entire repeater chain was analog and was
not distributed with VoIP, there was no latency involved in mobile-to-mobile or
mobile-to-portable transmissions. Radio traffic was at times very, very dense and
contentious; and the system spent significant amounts of time in nearly constant
multi-channel keyup handling the message traffic. Computer Network and Telco All outside network connections such as Dedicated
Internet Access, Data circuits to various points in the Of those landline T1s, #1 was 24 Telco
dialtone circuits, #2 was a Class C Dedicated Internet Access circuit, and the remainder
were involved with the various incoming data streams for tracking from Verizon AirCard,
Globalstar, Inmarsat, and other sources. The Telco T-1 was wired to DARPA Contractor CenGen Networking, Inc. handled the complex data circuits
and routing equipment for tracking and computer networking, and provided integration for
the several vehicle tracking methods into a single data stream. That data stream was
routed onto the local LAN in Primm along with the VoIP packet stream for the Telex/Vega
dispatch equipment. A near-full-motion video stream from the Stoddard Starting Line as
well as a video stream from the CenGen then bridged the Primm and Stoddard LANs over 3rd St. R & Ds microwave network. Of the four T1s available on the microwave network, #1 was allocated to 24 E&M circuits for radio system interconnect, #2 was allocated to Telco transport from Primm to Stoddard, #3 was allocated to an Ethernet Bridge between Primm and Stoddard, and #4 was allocated for Tracking Data. The combined Radio, Microwave, and Data
Communications system worked flawlessly together. Tear-Down of the systems began the day after the event on March 14, 2004 and was completed and the last truck left on March 18, 2004 -- 4 days from fully operational to totally clear of the area.
Summary 3rd St. R & D's unique experiences with this project have given us an even greater range of insights into techniques and methods which have application in not only transportable, temporary systems; but also in fixed systems which are required to cover large areas. Our systems have application not only in special event communications, but also in disaster restoration and other mission critical temporary situations. Our existing experience base with large systems is now vastly expanded to include a variety of networking technologies and equipment. Consultation and installation services involving voting systems, Vega Dispatch Products, and Adtran Wireless and Networking products are available. These technologies and techniques can be easily and cost-effectively combined with or added to existing fixed systems to improve their performance and coverage. Contact: Greg Carttar 417-336-4045
The Vendors: Motorola - What else can you say? People we've met on this project that are just the best: Cengen - Certified Cisco
Networking Consultants Special thanks to: Phil Angell, Platronics Communications The WHOLE staff at Primm Valley Resorts, Primm, Nevada |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||
back to main or hit your "back" button |