17-03-2014, 06:42 PM
Abstract—In this paper, a novel demodulation structure in
high-frequency (HF) channel is introduced and studied.
Frequency-offset estimation, fractionally spaced adaptive
equalization and carrier recovery techniques are involved.
Frequency-offset correction is carried out based on channel
impulse response estimation. Fractionally spaced adaptive
equalization and carrier recovery are combined with nonlinear
decision feedback loop structure? and the initial value
of tap coefficients of equalization are obtained from the
previous channel estimation. Simulation is conducted under
channel parameters with respect to MIL-STD-188-110B using
a Monte Carlo technique. The results show this novel
demodulation structure is feasible and efficient in HF channel,
the symbol error ratio can reach to 10−3 in 12dB SNR.
Keywords-HF channel; channel estimation; frequency-offset
estimation; fractionally spaced adaptive equalization; carrier
recovery
I. INTRODUCTION
During the last two decades, transmission of digital
information over HF channel has received continually
growing attention. The main reasons for this are that the use
of ionospheric propagation allows communication over
large distances and that the equipment is simple and
inexpensive. But the HF channel has disadvantages, it often
exists multipath fading which leads to channel impulse
response spanning several symbol intervals thus giving rise
to severe intersymbol interference(ISI), and another affect to
leads to signal distortion is frequency shifting due to
Doppler offset. Moreover, the differences in carrier
frequency at transmitter and receiver also causes frequency
shifting which can be detrimental to system performance.
Receiver performance is largely governed by frequencyoffset
estimation and equalization. Conventional dataaided(
DA) frequency-offset estimate schemes which work
well in AWGN channel such as Fitz[1], M&M [2] have bad
performance in HF channel due to the multipath fading. In
order to obtain the accurate and fast estimation, a robust
scheme according to the channel character is required to
implement. In this paper, frequency-offset correction is
carried out based on estimating the channel impulse response
at two or more instants not too far from each other during the
transmission of a known sequence of data. The sample points
are selected relatively close so any changes in the channel
can be assumed negligible except for any rotation due to
frequency offset. HF channel exists multipath fading which
leads to severe intersymbol interference, DFE[3] is a long
established attractive equalization technique for such channel
exhibiting deep in-band nulls due to selective fading, the
demodulation structure here adopts Kalman adaptive filters[4]
scheme to achieve adjustment of DFE coefficients and uses
square-root techniques to derive efficient and stable versions.
Moreover, symbol spaced equalizer requires the timingrecovery
module working on an optimal state, which is
difficult to achieve in practice, thus here use fractionally
spaced equalizer[5] to eliminate ISI and timing-error
together. The carrier recovery algorithm is based on digital
phase-locked loop(DPLL) and combined with adaptive
equalization to constitute a joint structure which can
effectively eliminate phase error and channel distortion.
Section 2 gives a detailed description of the joint scheme.
The rest of the paper is organized as follow. The simulation
performance of the structure is presented in section 3, and
finally, section 4 summarizes the results of the paper.
high-frequency (HF) channel is introduced and studied.
Frequency-offset estimation, fractionally spaced adaptive
equalization and carrier recovery techniques are involved.
Frequency-offset correction is carried out based on channel
impulse response estimation. Fractionally spaced adaptive
equalization and carrier recovery are combined with nonlinear
decision feedback loop structure? and the initial value
of tap coefficients of equalization are obtained from the
previous channel estimation. Simulation is conducted under
channel parameters with respect to MIL-STD-188-110B using
a Monte Carlo technique. The results show this novel
demodulation structure is feasible and efficient in HF channel,
the symbol error ratio can reach to 10−3 in 12dB SNR.
Keywords-HF channel; channel estimation; frequency-offset
estimation; fractionally spaced adaptive equalization; carrier
recovery
I. INTRODUCTION
During the last two decades, transmission of digital
information over HF channel has received continually
growing attention. The main reasons for this are that the use
of ionospheric propagation allows communication over
large distances and that the equipment is simple and
inexpensive. But the HF channel has disadvantages, it often
exists multipath fading which leads to channel impulse
response spanning several symbol intervals thus giving rise
to severe intersymbol interference(ISI), and another affect to
leads to signal distortion is frequency shifting due to
Doppler offset. Moreover, the differences in carrier
frequency at transmitter and receiver also causes frequency
shifting which can be detrimental to system performance.
Receiver performance is largely governed by frequencyoffset
estimation and equalization. Conventional dataaided(
DA) frequency-offset estimate schemes which work
well in AWGN channel such as Fitz[1], M&M [2] have bad
performance in HF channel due to the multipath fading. In
order to obtain the accurate and fast estimation, a robust
scheme according to the channel character is required to
implement. In this paper, frequency-offset correction is
carried out based on estimating the channel impulse response
at two or more instants not too far from each other during the
transmission of a known sequence of data. The sample points
are selected relatively close so any changes in the channel
can be assumed negligible except for any rotation due to
frequency offset. HF channel exists multipath fading which
leads to severe intersymbol interference, DFE[3] is a long
established attractive equalization technique for such channel
exhibiting deep in-band nulls due to selective fading, the
demodulation structure here adopts Kalman adaptive filters[4]
scheme to achieve adjustment of DFE coefficients and uses
square-root techniques to derive efficient and stable versions.
Moreover, symbol spaced equalizer requires the timingrecovery
module working on an optimal state, which is
difficult to achieve in practice, thus here use fractionally
spaced equalizer[5] to eliminate ISI and timing-error
together. The carrier recovery algorithm is based on digital
phase-locked loop(DPLL) and combined with adaptive
equalization to constitute a joint structure which can
effectively eliminate phase error and channel distortion.
Section 2 gives a detailed description of the joint scheme.
The rest of the paper is organized as follow. The simulation
performance of the structure is presented in section 3, and
finally, section 4 summarizes the results of the paper.