RE: stds-80220-ch-models: New channel measurements
Ken,
Thank you for planning some very exciting experiments! Definitely the measurements
will help us better understand MIMO implications for system capacity and phy-mac design.
My understanding is that it's planned to use time-shifted copies of
a prototype PN sequence as sounding signals, and 16 transmitters at 24 dBm
are used. The timing difference between neighboring time-shifted copies
is 3.2 microseconds. Given that in downtown area signficant multipaths
can be encountered, I suggest the timing difference be increased, e.g to 25.6 microseconds
(ITU vehicular model B's excess delay is 25 microseconds). One solution is
to increase the PN sequence length by 8 times, then it will take 8x51.1 (400) microseconds,
and the timing difference between time-shifted neighboring copies is 25.6 microseconds.
With an increased PN sequence length, the channel estimation could also improved.
(The area where channels can be reliably sounded can be found from cell planning tools.
It's possible 24 dBm output is not strong enough, then 9 dB more gain can become useful).
To capture Doppler fading, the setup with 4 TX antennas & 4 RX antennas seems a very good choice.
In the plan, the BTS antennas will be put at a height of 17m. One may also want to put
those antennas at the top of buildings of different height in the downtown area to sound channels.
As BTS antennas are likely to be installed at the top of buildings.
It would be also interesting to have different spacings for BTS antennas, e.g. one setup with
BTS antennas separated by half wavelength, another setup with BTS antennas separated by several wavelengths,
yet another by tens of wave-lengths.
Regards,
Weidong Yang,
Navini Networks
-----Original Message-----
From: Ken Allen [mailto:allen@its.bldrdoc.gov]
Sent: Friday, January 16, 2004 3:02 PM
To: stds-80220-ch-models@ieee.org
Cc: Frank Sanders; Peter Papazian; John Lemmon
Subject: stds-80220-ch-models: New channel measurements
Dear All,
I want to describe the measurement capability and measurements that we
are planning to do this year in support of MIMO technology development.
INTRODUCTION
As part of a program to support the development of spectrally efficient
technologies, in particular, MIMO technology, the Institute for
Telecommunication Sciences (ITS) is planning to make mobile, MIMO
channel measurements in 2004. To maximize the value of these
measurements, ITS will accept input from IEEE 802.20 participants and
other interested parties regarding how these measurements should be
made to provide results useful in the development of the 802.20
standards and mobile, broadband data communications in general.
ITS is a Federal research laboratory housed in the National
Telecommunications and Information Administration and located in
Boulder, Colorado. ITS is a central resource for telecommunications
research for Government and industry and has made major contributions
to propagation modeling and measurements since the second world
war. Experienced, internationally recognized experts in radio channel
modeling and measurements as well as state-of-the-art equipment will be
utilized in this research effort.
MEASUREMENT CAPABILITIES
As currently configured, the measurement system transmits up to 16
signals at 2250 MHz, which are received on up to 4 receivers. By using
BPSK modulation and pseudo-random noise (PN) binary codes, the
complex-valued impulse response of each element in the transmission
matrix can be measured. This is done by digitally sampling the IF of
each receiver and cross correlating the samples with a sampled version
of each of the transmitted signals.
Currently each transmitter has an output power of 24 dBm. A single
511-bit PN code word is used at equally spaced delays for each transmit
channel. The code word is clocked at 10 MHz. Thus, it takes 51.1
microseconds to transmit the code word, which then repeats. Each
channel’s code word is delayed 3.2 microseconds from the previous
channel limiting the time delay measurement range to 3.2
microseconds. The transmitted spectrum falls off from the carrier
frequency as the sinc function with the first nulls on both sides
separated by 20 MHz. The receivers sample at 40 MHz with 14-bit
resolution.
The configuration of this system can be changed, if necessary, in a
number of ways, including.
1. A different carrier frequency could be used.
2. The number of received channels can be increased beyond 4.
3. Longer or shorter PN code words can be used.
4. OFDM signals can be transmitted instead of a single BPSK carrier.
5. The code word can be clocked at up to 20 MHz.
Data collection is currently done in two modes. In the burst mode, the
received signal is sampled for the duration of one code word. This is
repeated up to 256 times with a spacing of from 102.2 microseconds to
many milliseconds. The spacing determines the maximum Doppler frequency
that can be unambiguously measured. A burst measurement can be recorded
each second or spaced further apart.
In the continuous measurement mode, all receiver signals are sampled
for the duration of one code word. This is repeated continuously as
fast as every 400 microseconds, until the disk is full (30 GB).
CURRENT MEASUREMENT PLAN
The current plan for the measurements this year is to transmit at 2250
MHz from 4 base-station like antennas, each BPSK modulated at a 10 MHz
chip rate using one repeating, 511-bit, PN code word, with each of the
four signals separated by a delay of 128 bits or 12.8 microseconds.
Four receiving antennas representative of the type that may be used on
a vehicle and of the type that may be used on a handheld device will be
used at the mobile terminal. The signal from each of the 4 receiving
antennas will be digitally sampled at 40 M-samples/second (14-bit
resolution) using the burst-sampling mode described above. The spacing
between the waveform samples will be chosen to provide better than a
Nyquist-frequency sampling rate for the maximum possible Doppler
frequency during the measurement.
Outdoor measurements will be made in downtown Denver, an urban
area. (Subsequent measurements may be made with the antennas in several
different physical configurations. This would enable the measurement of
MIMO capacity for various antenna diversity schemes involving various
antenna spacings and polarizations.)
The outdoor measurements will utilize a fixed base station with a
height of 17m and a mobile measurement van. (Future measurements may
be made indoors or outdoor-to-indoor.
The currently planned processing of the measurement data is as
follows. Each set of MIMO waveform samples (51.1 microseconds long),
will be cross-correlated in post processing with a digitally sampled
version of the transmitted signal producing the complex values (phase
and amplitude) of the elements of the MIMO transmission matrix. The
rank of each matrix will be computed. (Note: This can be done using a
variety of combinations of numbers of transmit and receive antennas.)
The statistics of the rank and its time dynamics (such as coherency
time and distance) will be determined for each combination of path
type, environment, and antenna diversity.
INPUTS WELCOME
Suggestions are welcome from individuals, and the 802.20 channel
modeling group as a whole, of modifications to the measurement system,
measurement plan, and data analysis that will provide more useful
results. Unfortunately, I was not able to attend the Vancouver meeting,
but I plan to attend the March meeting.
To save time, I will respond to questions and/or suggestions at any
time before the next meeting. I can be reached at 303-497-5474 or
kallen@its.bldrdoc.gov.
Thank you very much for your help.
Ken
*********************************************************
Kenneth C. Allen *
Wireless Program Leader *
NTIA/Institute for Telecommunication Sciences *
325 Broadway *
Boulder, CO 80305 *
*
Phone: 303-497-5474 *
Fax: 303-497-3680 *
E-mail: kallen@its.bldrdoc.gov *
*********************************************************