 |
Description  |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to the field of ultra wideband communications and
radar. More particularly, it relates to reliable reception and processing
of ultra wideband electromagnetic pulses in the presence of noise, strong
interference and/or jamming.
2. Background of Related Art
One of the first ultra wideband (previously referred to as baseband,
carrier-free or short pulse) receivers was patented in 1972 by Ken Robbins
while at the Sperry Research Center, U.S. Pat. No. 3,662,316. This
receiver utilized a "dispersionless" broadband transmission line antenna
together with a biased tunnel diode located in the transmission line for
detecting the total energy in a pulse and expanding the resultant output
in the time domain so that conventional, lower speed circuitry may be used
for processing. The tunnel diode was biased to operate as a monostable
multivibrator as disclosed in 1962 in Gentile, S. P., Basic Theory and
Application of Tunnel Diodes, Van Nostrand, N.J., ch. 8 "Pulse and
Switching Circuits" (1962). The receiver took advantage of the tunnel
diode's unique characteristic of changing state when the area under the
current vs. time envelope, i.e., the charge carriers passing through the
device, exceeded a prescribed number of picocoulombs. This change in state
yielded a recognizable, detectable event or output voltage. Sperry's
tunnel diode detector (TDD) receiver was used in a number of applications
including baseband communications, liquid level sensing, object detection
and radar. It was soon observed, however, that the Robbins TDD was subject
to operating point bias drift due to temperature and power supply
fluctuations. This bias drift impacted negatively the system's overall
sensitivity and increased the false alarm rate.
In 1976, Nicolson and Mara introduced a constant false alarm rate (CFAR)
circuit to the tunnel diode detector receiver that is described in U.S.
Pat. No. 3,983,422. The CFAR circuit employed a logic circuit that sampled
noise dwells and data dwells to dynamically adjust a variable threshold of
the tunnel diode. This feedback circuit operated in such a manner that the
false alarm rate, as measured by the number of hits received due solely to
noise during a fixed time interval, was held constant regardless of
temperature fluctuations, power supply voltage changes, device aging, etc.
The CFAR receiver was utilized in the development of baseband speed
sensing, collision avoidance, and radar docking prototypes.
In 1987, an anti-jam circuit was introduced into the CFAR receiver. This is
described in U.S. Pat. No. 4,688,041. Since the baseband receiver was
extremely broadband, with typical bandwidths of hundreds of MHz to GHz, it
was found to be extremely susceptible to in-band interference and jamming
since the tunnel diode circuit could not distinguish between valid and
unwanted signals. Such in-band signals caused a significant reduction in
receiver sensitivity by causing the CFAR loop to back-off the sensitivity
of the tunnel diode detector. The anti-jam circuit disclosed in U.S. Pat.
No. 4,688,041 used the jamming signal itself (if sufficiently strong), or
else an internally switched continuous wave (CW) signal, as a local
oscillator signal to heterodyne the incoming signal prior to detection.
However, this anti-jam circuit proved to be ineffective in the presence of
barrage (broadband) noise jamming or interference, and/or multiple in-band
CW interfences. In the case of barrage noise, no reference frequency is
provided by the interference with which to down convert the incoming
signal, and the system reverts to single-conversion superheterodyne
operation with an internal first local oscillator. The broadband noise is
also down converted with the signal, and no anti-jam improvement is
obtained. In the latter case of multiple in-band CW interferers, the
circuitry will use one of these tones, or a linear combination depending
upon the third order intercept properties of the design. In this case, the
remaining tones are also heterodyned to near baseband and act once again
as strong in-band jamming signals.
Also in 1987, U.S. Pat. No. 4,695,752 disclosed a narrow range gate added
to the existing baseband CFAR receiver. The reduction in range gate size
had the effect of reducing unwanted noise and interference by more closely
matching the detector with the received pulse duration. The inventor of
this patent purports to achieve nanosecond range gate intervals through
the use of two Germanium (Ge) and a single Gallium Arsenide (GaAs) tunnel
diode.
In 1994, U.S. Pat. No. 5,337,054 to Ross and Mara disclosed a coherent
processing tunnel diode UWB receiver. These inventors claim to have
improved tunnel diode detector receiver sensitivity by using a tunnel
diode envelope generator to perform a superheterodyne conversion whereby
the available charge for triggering the tunnel diode is maximized. Ross
and Mara considered only single pulse ultra wideband detectors; i.e.,
detectors which make a binary, or hard, decision (Logic 1 or Logic 0) at
every sampling instant. However, their patent discloses a sliding average
of detector hits, noise dwell or data dwell, in any group of thirty-two
consecutive periods (col. 4, lines 35-39). Averaging of all hits,
including data dwells, provides an average of the noise dwells which is
skewed because of the inclusion of the data dwells. Moreover, to reduce
the effects of the skewing, a large number of noise dwells must be
detected for each data dwell detected, ultimately reducing data rates.
There have been other patented UWB receiver designs in which a multiplicity
of pulses (typically several thousand) are first coherently added, or
integrated, before a binary (bit) decision is made (e.g., U.S. Pat. Nos.
5,523,760; 4,979,186; and 5,363,108). The UWB detectors of the present
invention do not require coherent addition of a multiplicity of pulses,
but rather have sufficient sensitivity to operate on a single pulse basis.
Only false alarm rate is typically computed by previous UWB
receiver/processor designs, and thus the system bit error rate (BER), and
accordingly the receiver operating characteristic (ROC) are unknown. In
practice, the tunnel diode bias is "backed off" from the CFAR level to
reduce the BER to an acceptable level. Unfortunately, since the BER is a
very sensitive function of the tunnel diode bias level, this can result in
a significant reduction in receiver sensitivity to achieve a desired BER.
As disclosed in U.S. Pat. No. 3,662,316, in a tunnel diode UWB receiver,
the tunnel diode changes state whenever the accumulated charge on the
device exceeds a given threshold. Mathematically, the performance of the
tunnel diode detector in additive white Gaussian noise (AWGN) can be
described by the following set of equations:
##EQU1##
where P.sub.d is the probability of detection, P.sub.fa is the probability
of false alarm, s(u) is the received UWB waveform, n.sub.w (u) is additive
white Gaussian noise with double-sided power spectral density N.sub.0 B, B
is the detection signal bandwidth, T is the diode dwell sensitivity
interval, and T.sub.h is a threshold value.
While previous designs of the CFAR tunnel diode receiver have functioned
reliably as an ultra wideband single pulse detector, their use in modern
communication and radar applications have presented numerous drawbacks:
1. The prior art designs remain susceptible to in-band interference and
jamming, particularly broadband or barrage noise jamming and multiple CW
interferers.
2. The requirement to continuously adjust bias to the tunnel detector to
maintain a given constant false alarm rate (CFAR) conventionally requires
a minimum number of noise dwells to take place for each data
dwell--typically thirty-two or more noise dwells for each data dwell--to
achieve false alarm rates less than a few percent. This severely restricts
the maximum data rate at which a single detector can operate since data
and noise dwells must operate at different time intervals. In addition,
the speed at which the tunnel diode detector can respond to sudden changes
in the electromagnetic environment is limited. Hence, impulsive noise
(which is nearly always present) can create burst errors in the data
stream, corrupting data integrity.
3. Receiver sensitivity is conventionally backed-off to achieve a desired
BER, providing an UWB receiver which has reduced distance capability and
slower data rates.
SUMMARY OF THE INVENTION
A microwave tunnel diode is utilized as a single pulse detector for short
pulse, impulse, baseband or ultra wideband signals. The UWB receiver has a
number of unique features which permit highly sensitive operation at
extremely high speeds (multiple Mb/s) with high immunity to in-band
jamming.
For instance, the tunnel diode detector bias point is preferably determined
only once, and preferably at system start-up, through an automatic
calibration procedure. In this fashion, the tunnel diode detector is set
to its highest sensitivity point relative to the desired bit error rate
performance based upon internal noise only, and remains at that point
during the entire reception process. Conventional CFAR-based UWB receivers
continually update the detector bias point, resulting in reduced detector
sensitivity in the presence of in-band jamming (i.e., receiver back-off),
and extremely slow response times because of the need to constantly
recalculate the false alarm rate. Rather than adjust the bias to the
tunnel diode detector, the present invention adjusts the attenuation of
the incoming UWB signal. An adaptive dynamic range extension process using
a high speed, Gallium Arsenide (GaAs) voltage variable attenuator (VVA)
provides high noise immunity. The instantaneous attenuation level is
determined by periodically sampling the ambient noise.
A high speed switch time-gates the tunnel diode detector by switching
between a gate active mode in which the tunnel diode detector is connected
to the receiver front end circuitry for reception of an UWB pulse; and a
gate inactive mode in which the signal input is removed from the detector
and charge stored in the tunnel diode detector is discharged. Conventional
designs have previously used the tunnel diode bias voltage itself to gate
the tunnel diode detector, resulting in reduced RF sensitivity due to
transients induced by the bias switching circuitry.
It is therefore an object of the present invention to provide an UWB
receiver which operates with extremely high sensitivity at extremely high
speeds with high immunity to inband jamming.
It is a further object to provide a method of calibrating the bias
threshold of an UWB receiver.
It is another object to provide an UWB receiver which biases a tunnel diode
detector at its highest sensitivity point based on internal noise only.
It is a further object to provide an UWB receiver which uses a high speed,
adaptive dynamic range extension process.
It is also an object of the present invention to provide an UWB receiver
which switchably discharges the tunnel diode detector between data dwells.
An additional object is to provide an UWB receiver having separate data and
noise dwell measurement circuitry.
BRIEF DESCRIPTION OF THE DRAWINGS
These and other objects will become readily apparent to those of ordinary
skill in the art from the detailed description of the presently preferred
exemplary embodiments with reference to the drawings, in which:
FIG. 1 is a schematic diagram of a high data rate UWB receiver according to
a first embodiment of the present invention.
FIG. 2 is a timing diagram showing the operation of switches S1 to S3 in
FIG. 1 with respect to an UWB received pulse, e.g., timing for two noise
dwells per data dwell.
FIG. 3 is a flow diagram for the tunnel diode detector UWB receiver
according to the present invention.
FIG. 4 is a photograph of a circuit board of a transceiver utilizing the
present invention.
FIG. 5 is a schematic diagram of an ultra high data rate UWB receiver
according to a second embodiment of the present invention.
FIG. 6 is a timing diagram of the circuit of FIG. 5.
FIGS. 7A to 7C show the current-to-voltage (I-V) characteristics of a
tunnel diode (FIG. 7A), a Schottky Diode (FIG. 7B) and a back tunnel diode
(FIG. 7C), respectively.
DETAILED DESCRIPTION OF ILLUSTRATIVE EMBODIMENTS
The tunnel diode detector circuitry and process steps of the present
invention provide an UWB receiver which is highly immune to many forms of
in-band jamming; can operate at extremely high data rates (tens to
hundreds of megabits per second ›Mb/s!; and provides a high speed
threshold which compensates for the level of noise and guarantees a
desired ROC performance. A TDD UWB receiver using this technique provides
single pulse detection of a 200 .mu.W (average power) UWB signal at
distances exceeding 50 miles.
The tunnel diode(s) of the UWB receiver is (are) biased to operate near its
peak current value I.sub.p, which occurs just prior to the tunnel diode's
AC negative resistance region as shown in FIG. 7A. This is in sharp
contrast to the operation of a conventional microwave Schottky diode
detector as shown in FIG. 7B, or a conventional back tunnel diode detector
as shown in FIG. 7C, which rely on signal rectification with low or zero
offset voltages for the detection of low level microwave signals.
The tunnel diode detector integrates directly the received voltage (current
times input impedance) over the dwell interval. Since the integral of
white Gaussian noise is equivalent to Brownian motion (also called a
Wiener process after the mathematician Norbert Wiener), the tunnel diode
detector transforms noise statistics in such a manner that the smaller the
dwell interval, the larger the available signal-to-noise ratio. (Gikhman,
I. I. et al. Introduction to the Theory of Random Processes, Dover
Publications, New York (1969).
The detector according to the present invention achieves high processing
gain without resorting to the coherent addition of multiple pulses
required by other designs such as those disclosed by U.S. Pat. Nos.
5,523,760, 4,979,186, and 5,363,108. Moreover, unlike previous receiver
designs that were limited to data rates of approximately 20 Kb/s
(kilobit/second) for voice and data communications as well as radar
applications, the UWB receiver according to the present invention is
capable of data rates in the multiple Mb/s (megabit/second) range. This
high speed operation enables full duplex or two-way transmission and
reception of stereo quality voice, video imagery and other data rate
critical applications.
FIG. 1 shows a simplified schematic diagram of a first embodiment of a high
speed UWB receiver according to the present invention. In FIG. 1, a
received UWB pulse is input, through calibration switch S1, to a wideband,
high gain RF amplifier 100. Calibration switch S1 is used to switch
between an operational mode with switch S1 in the down position as shown
in FIG. 1, and a calibration mode with switch S1 in the up position
connecting the input of high gain RF amplifier 100 to ground through
impedance matching resistor 108, which in this embodiment is 50 ohms. In
the preferred mode, calibration switch S1 is switched only once after
power-up and is activated by a programmable logic device or microprocessor
110.
Receiver `Operational Mode`
With calibration switch S1 in the operation mode position, after RF
amplification, the UWB pulse is input to voltage variable attenuator (VVA)
102, whose primary function is to provide high speed, adaptive dynamic
range extension. The adaptive dynamic range extension process is
particularly useful in the presence of in-band interference and noise.
According to the process, the attenuation of VVA 102 is adjusted, under
microprocessor control, by the output of a digital-to-analog converter
(DAC) 104. A ten bit DAC is sufficient for the purposes of DAC 104,
although a 12 bit DAC is preferred to provide a greater margin for error.
The DAC 104 settling time should be fast enough to provide adequate and
timely control of the VVA 102 before the next gate change. As an example,
for a 10 Mb/s data rate, the DAC 104 settling time should be less than
approximately 100 ns. A slower DAC 104 can be used if some hysteresis is
tolerable in the VVA 102 control setting.
The UWB signal is attenuated based on an instantaneous measurement of
noise, allowing the tunnel diode detector to be maintained at its peak
current value to provide maximum sensitivity, rather than the conventional
method of backing-off the bias current of the tunnel diode detector based
on the presence of noise. The particular VVA 102 used provides a 60 dB
attenuation range, and is ideally linear. Significant non-linearities in
the VVA 102 are compensated in the microprocessor 110 through
corresponding control of DAC 104. To obtain a desired range of
attenuation, multiple VVAs may be cascaded. For instance, two 30 dB VVAs
may be cascaded to obtain 60 dB range of attenuation, or two 40 dB VVAs
may be cascaded to obtain 80 dB range of attenuation.
Alternatively, a digitally-controlled microwave step attenuator may be
substituted for the VVA 102. A step attenuator implements a set of
discrete attenuation levels versus the continuum of levels achievable with
VVA 102. Thus, it would have somewhat decreased performance in the
presence of jamming and impulsive noise due to quantizing effects in
achieving the desired operating bias. Although 2 dB increments have been
implemented, the difference in the attenuation between levels of the step
attenuator is preferably 0.5 dB or less.
The amplified, and VVA-attenuated, UWB pulse is subsequently fed via
capacitor 106 to a detector gate formed by switch sections S2a, S2b and
resistors 118, 120. Switch sections S2a and S2b act in tandem to
alternatively apply and remove the conditioned microwave UWB pulse from
the input terminal of a tunnel diode detector 122. When switch sections
S2a and S2b are both in the DOWN position (as shown in FIG. 1), the UWB
pulse is applied to tunnel diode detector 122. When both switch sections
S2a, S2b are in the UP position, resistor 118 terminates the output of the
amplifier 100/VVA 102 chain to prevent any potential instabilities due to
a standing wave ratio (SWR) mismatch. At the same time, switch S2b shorts
resistor 120 across tunnel diode detector 122 to remove any stored charge.
The particular tunnel diode detector 122 used was obtained from GERMANIUM
POWER DEVICES CORP. in Andover, Mass., Model No. TD272A.
Resistor 120 used with the disclosed embodiment has a value of 10 ohms
which is selected to be sufficiently low to reset the tunnel diode
detector 122 without causing significant current unbalance in the constant
current source. The power source Vcc was 3.3 volts, although any
conventional voltage supply level may be implemented with appropriate
circuit component families.
A voltage-controlled constant current source 142 controls the bias on
tunnel diode 122. Constant current source 142 includes transistor 130,
resistors 124, 132 and 134, filtering capacitor 128, and protection diode
136. Transistor 130 is a Model No. 2N3906, resistor 124 has a value of 100
ohms, and filtering capacitor 128 has a capacity of 33 picofarads.
Switch S3 selects the voltage control mode of constant current source 142.
In the UP position as shown in FIG. 1, switch S3 selects a `Data
Threshold` voltage control mode which sets the tunnel diode bias of the
tunnel diode detector 122 for data detection during data dwells. The `Data
Threshold` mode is used to set the bias of tunnel diode detector 122 to
the correct sensitivity portion of its voltage-current (V-I)
characteristics for the detection of an UWB pulse, based on the desired
BER. The data threshold setting circuit comprises resistor 138 and
digital-to-analog converter (DAC) 112. Switches S2 and S3 are controlled
by a high speed programmable logic device (not shown) which can either
have its own oscillator or clock, or share an oscillator or clock with
microprocessor 110.
Ten-bit DACs are sufficient for DACs 112 and 114, though twelve bit DACs
provide a greater margin for error. DACs 112 and 114 are preferably set
only once, during a calibration routine, and are not changed thereafter.
The particular DACs 112, 114 used in the disclosed embodiments are LINEAR
TECHNOLOGIES, Model No. LTC1453, which have a serial input. Of course,
parallel input DACs may be implemented instead of serial input DACs.
In the DOWN position, switch S3 selects a `Noise Threshold` voltage control
mode which utilizes resistor 140 and DAC 114. The `Noise Threshold` mode
is used to set the bias of tunnel diode detector 122 to the correct
sensitivity portion of its V-I characteristics for measuring either
internal or external noise power. Switches S1, S2 and S3 are preferably
suitably fast semiconductor switches, e.g., Gallium Arsenide (GaAs)
semiconductor switches having low insertion loss at microwave frequencies.
The particular switches used are available from MINI CIRCUITS, Model No.
YSW-2-50dR.
During a dwell, when sufficient charge accumulates at the terminals of
tunnel diode detector 122, it changes state thereby causing a voltage
swing to occur across its terminals. Resistor 126 couples this voltage
swing to a comparator latch 116 which includes comparator 116a, shown
separately in FIG. 1 for clarity. A separate comparator could
alternatively be implemented. A reference voltage source 160 applied to
the negative input of the comparator is set to about 250 millivolts so
that latch 116 detects and latches voltages above this reference voltage.
Latch 116 generates a digital CMOS-compatible voltage signal for
subsequent processing by a microprocessor 110. Any logic family may be
utilized using appropriate output levels, e.g., TTL, ECL, etc. The
particular comparator latch 116 used is a Model No. AD9696.
A noise dwell is defined to be the state in which the detector gate formed
by switch S2 shown in FIG. 1 is enabled, i.e., in the DOWN position, and
the threshold select switch S3 is in the `Noise Mode` or DOWN position to
select the Noise Threshold DAC 114. A Data Dwell is defined as the state
in which switch S2 is in the enabled or DOWN position, and the threshold
select switch S3 is in the `Data Mode` or UP position to select the Data
Threshold DAC 112. The frequency of data dwells is related to the
signaling rate. For maximum speed, the number of noise dwells per data
dwell is minimized, but in the preferred embodiment at least two noise
dwells are required for each data dwell. In lower speed applications,
accuracy can be improved by performing an i | | |
