def receiver_nonlinear_noise(name):
import joblib
from library.electrical import Resampler
import resampy
from library import matched_filter
signal = __reconstruct_signal(name)
fibers = [LinearFiber(0.2, 16.7, 80, 0, 1550) for _ in range(12)]
signal = cd_compensation(fibers, signal, signal.fs_in_fiber)
signal.samples = resampy.resample(signal[:],
signal.sps_in_fiber,
2,
axis=-1,
filter='kaiser_fast')
signal = matched_filter(signal, 0.02)
signal.samples = signal[:, ::2]
phase = np.angle(
np.mean(signal.samples / signal.symbol, axis=-1, keepdims=True))
signal.samples = signal.samples * np.exp(-1j * phase)
signal.inplace_normalise()
noise = signal.samples - signal.symbol
noise_power = np.sum(np.mean(np.abs(noise)**2, axis=-1))
nonlinear_snr = 10 * np.log10((2 - noise_power) / noise_power)
return nonlinear_snr, signal, noise
def receiver_nonlinear_noise(name, config_ith):
import joblib
from library.electrical import Resampler
import resampy
from library import matched_filter
config = joblib.load('../simulate_data/config_setting')[config_ith]
signal = __reconstruct_signal(name,config)
fibers = fiber_for_reconstruct(config)
signal = cd_compensation(fibers,signal, signal.fs_in_fiber)
signal.samples = resampy.resample(
signal[:], signal.sps_in_fiber, 2, axis=-1, filter='kaiser_fast')
signal = matched_filter(signal, 0.02)
signal.samples = signal[:, ::2]
phase = np.angle(np.mean(signal.samples / signal.symbol, axis=-1, keepdims=True))
signal.samples = signal.samples * np.exp(-1j*phase)
signal.inplace_normalise()
noise = signal.samples - signal.symbol
noise_power = np.sum(np.mean(np.abs(noise) ** 2, axis=-1))
nonlinear_snr = 10 * np.log10((2 - noise_power) / noise_power)
return nonlinear_snr,signal
def extract_spectrum(name, config_ith):
import joblib
from library.electrical import Resampler
import resampy
from library import matched_filter
from scipy.signal import welch
config = joblib.load('../simulate_data/config_setting')[config_ith]
signal = QamSignal.load(name,True)
fibers = fiber_for_reconstruct(config)
signal = cd_compensation(fibers,signal, signal.fs_in_fiber)
signal.samples = resampy.resample(
signal[:], signal.sps_in_fiber, 2, axis=-1, filter='kaiser_fast')
signal = matched_filter(signal, 0.02)
[f,pxx] = welch(signal.samples[0],detrend=None,return_onesided = False,nfft = 1024,fs=signal.fs)
signal.inplace_normalise()
from library.receiver_dsp import LMS_PLL
from library.receiver_dsp import syncsignal_tx2rx
lms = LMS_PLL(0.1, 321, 3, 3, 0.0001)
signal = lms.equalize(signal)
tx_symbol = syncsignal_tx2rx(signal.samples, signal.symbol)[:,:signal.samples.shape[1]]
noise = signal.samples - tx_symbol
noise_power = np.sum(np.mean(np.abs(noise) ** 2, axis=-1))
# pxx,lms,signal.samples,tx_symbol
return f,pxx, lms, signal, tx_symbol
def prop(self, signal, span, roll_off):
from library import cd_compensation, matched_filter, LMS
from library.receiver_dsp import syncsignal_tx2rx
import resampy
signal = cd_compensation(span, signal, signal.fs_in_fiber)
signal.samples = resampy.resample(signal.samples, signal.sps_in_fiber,
signal.sps)
signal = matched_filter(signal, roll_off)
angle = np.angle(
np.mean(signal[:, ::2] / signal.symbol, axis=-1, keepdims=True))
signal.samples = signal.samples * np.exp(-1j * angle)
signal.inplace_normalise()
lms = LMS(321, 3, 3, 0.001)
signal = lms.equalize(signal)
signal.symbol = syncsignal_tx2rx(signal.samples,
signal.symbol)[:, :signal.shape[1]]
noise = signal.samples - signal.symbol
power = np.mean(np.abs(noise)**2, axis=-1)
power = np.sum(power)
snr = (2 - power) / power
snr = 10 * np.log10(snr)
print(snr)
self.equalizer = lms
return signal