def alg_2(csi):
"""alg_2
Note: N_p = 1
"""
p = np.power(np.abs(phy_ifft(csi, scidx(20, 2), axis=1)), 2.0)
# find_peak() is better
s = p[:, :20].argmax(axis=1).reshape(csi.shape[0], -1)
N_sto = scidx(20, 2)[mode(s, axis=1)[0][:, 0]]
N_sc = 64
csi = derotate_formual16(csi, -2 * np.pi * N_sto / N_sc)
return csi
def func_3(csidata):
"""CSI: time-phase"""
s_index = 15 # subcarrier index
csi = csidata.get_scaled_csi_sm()
t = csidata.timestamp_low / 1000000 - csidata.timestamp_low[0] / 1000000
phase = np.unwrap(np.angle(csi), axis=1)
phase = calib(phase, scidx(20, 2))
plt.figure()
plt.plot(t,
phase[:, s_index, 0, 0],
linewidth=0.3,
label='subcarrier_15_0_0')
plt.plot(t,
phase[:, s_index, 1, 0],
linewidth=0.3,
label='subcarrier_15_1_0')
plt.plot(t,
phase[:, s_index, 2, 0],
linewidth=0.3,
label='subcarrier_15_2_0')
plt.legend()
plt.title('csi-phase')
plt.xlabel('time(s)')
plt.ylabel('phase')
plt.show()
def fig_4(athdata, index=26):
"""fig_4
Ref:
[FFT Length (802.11n/ac/ax)](http://rfmw.em.keysight.com/wireless/helpfiles/89600B/WebHelp/Subsystems/wlan-mimo/Content/mimo_adv_fftsze.htm)
"""
csi = athdata.csi
# Fig.4(a)
phase = np.unwrap(np.angle(csi), axis=1)
phase = calib(phase, scidx(20, 1))
plt.subplot(3, 1, 1)
plt.plot(phase[index].reshape(csi.shape[1], -1))
# Fig.4(b)
amplidute = 10 * np.log10(np.abs(csi) + 1 / 10**6) # db
plt.subplot(3, 1, 2)
plt.plot(amplidute[index].reshape(csi.shape[1], -1))
# Fig.4(c)
# scipy.fftpack.fft or phy_fft ?
A = np.log10(np.abs(fft(csi, n=64, axis=1)))
plt.subplot(3, 1, 3)
plt.plot(A[index].reshape(64, -1), '*--')
plt.tight_layout()
plt.show()
def algshow(csidata):
s_index = scidx(20, 2)
csi = csidata.csi[:500]
plt.figure()
# alg_1
csi = alg_1(csi)
phase = np.unwrap(np.angle(csi), axis=1)
phase = calib(phase, s_index)
plt.subplot(3, 1, 1)
plt.plot(s_index, phase[200:300, :, 0, 0].T)
# alg_2
csi = alg_2(csi)
phase = np.unwrap(np.angle(csi), axis=1)
phase = calib(phase, s_index)
plt.subplot(3, 1, 2)
plt.plot(s_index, phase[200:300, :, 0, 0].T)
# alg_3
# csicopy = csi.copy()
# for i in range(100, 200):
# csicopy[i] = alg_3(csi, i)
# phase = np.unwrap(np.angle(csicopy), axis=1)
# phase = calib(phase, s_index)
# plt.subplot(3, 1, 3)
# plt.plot(s_index, phase[100:200, :, 0, 0].T)
plt.show()
def fig_6(csidata):
"""fig_6"""
csi = csidata.csi
s_index = scidx(20, 2)
phase = np.unwrap(np.angle(csi[:1000, :, 0, 0])).T
phase_diff = np.diff(phase)
phase_diff -= phase_diff.mean(axis=0)
phase_mean_0 = phase_diff.mean(axis=-1)
plt.figure()
plt.subplot(2, 1, 1)
plt.plot(s_index, phase_diff, 'y+', linewidth=0.3)
plt.plot(s_index, phase_mean_0, 'r-', linewidth=2.0)
patch_1 = mpatches.Patch(color='yellow', label=':1000_r0t0')
patch_2 = mpatches.Patch(color='red', label=':1000_r0t0_phase_mean_0')
plt.legend(handles=[patch_1, patch_2])
plt.title('fig6b: csi-phase_diff')
plt.xlabel('subcarriers')
plt.ylabel('phase_diff')
plt.subplot(2, 1, 2)
plt.hist(phase_diff[12],
bins=250,
histtype='stepfilled',
density=False,
color='red',
alpha=0.5)
plt.hist(phase_diff[13],
bins=250,
histtype='stepfilled',
density=False,
color='green',
alpha=0.5)
patch_1 = mpatches.Patch(color='red', label=':1000_s12r0t0')
patch_2 = mpatches.Patch(color='green', label=':1000_s13r0t0')
plt.legend(handles=[patch_1, patch_2])
plt.title('fig6c: csi-phase_diff')
plt.xlabel('phase_diff')
plt.ylabel('packets count')
plt.tight_layout()
plt.show()
def func_2(csidata):
"""CSI: subcarrier-amplitude(CFR)"""
csi = csidata.get_scaled_csi_sm()
amplitude = np.abs(csi)
s_index = scidx(20, 2)
plt.figure()
plt.plot(s_index, amplitude[:100, :, 0, 0].T, 'r-', linewidth=0.3)
plt.plot(s_index, amplitude[:100, :, 1, 0].T, 'g-', linewidth=0.3)
plt.plot(s_index, amplitude[:100, :, 2, 0].T, 'y-', linewidth=0.3)
patch_1 = mpatches.Patch(color='red', label=':100_r0t0')
patch_2 = mpatches.Patch(color='green', label=':100_r1t0')
patch_3 = mpatches.Patch(color='yellow', label=':100_r2t0')
plt.legend(handles=[patch_1, patch_2, patch_3])
plt.title('csi-amplitude')
plt.xlabel('subcarriers')
plt.ylabel('amplitude')
plt.show()
def func_4(csidata):
"""CSI: subcarrier-phase"""
csi = csidata.get_scaled_csi_sm()
s_index = scidx(20, 2)
phase = np.unwrap(np.angle(csi), axis=1)
phase = calib(phase, s_index)
plt.figure(4)
plt.plot(s_index, phase[:100, :, 0, 0].T, 'r-', linewidth=0.3)
plt.plot(s_index, phase[:100, :, 1, 0].T, 'g-', linewidth=0.3)
plt.plot(s_index, phase[:100, :, 2, 0].T, 'y-', linewidth=0.3)
patch_1 = mpatches.Patch(color='red', label=':100_r0t0')
patch_2 = mpatches.Patch(color='green', label=':100_r1t0')
patch_3 = mpatches.Patch(color='yellow', label=':100_r2t0')
plt.legend(handles=[patch_1, patch_2, patch_3])
plt.title('csi-phase')
plt.xlabel('subcarriers')
plt.ylabel('phase')
plt.show()
def func_6(csidata):
"""CSI: time-amplitude(CIR: OFDM symbol view)"""
csi = csidata.get_scaled_csi_sm()
s_index = scidx(20, 2)
amplitude1 = np.abs(phy_ifft(csi[:100, :, 0, 0], s_index, axis=1)).T
amplitude2 = np.abs(phy_ifft(csi[:100, :, 1, 0], s_index, axis=1)).T
amplitude3 = np.abs(phy_ifft(csi[:100, :, 2, 0], s_index, axis=1)).T
t = np.linspace(0, 64, 64)
plt.figure(6)
plt.plot(t, amplitude1, 'r-', linewidth=0.3)
plt.plot(t, amplitude2, 'g-', linewidth=0.3)
plt.plot(t, amplitude3, 'y-', linewidth=0.3)
patch_1 = mpatches.Patch(color='red', label=':100_r0t0')
patch_2 = mpatches.Patch(color='green', label=':100_r1t0')
patch_3 = mpatches.Patch(color='yellow', label=':100_r2t0')
plt.legend(handles=[patch_1, patch_2, patch_3])
plt.title('csi-CIR')
plt.xlabel('time(50ns)')
plt.ylabel('amplitude')
plt.show()
def fig_2(csidata, index=34):
"""fig_2
Note:
Fig. 2a uses data collected from a setup where transmitter and receiver arrays
directly face each other whereas, in Fig. 2b, the receiver is rotated by 90 degrees
Ref:
[Power Delay Profile](https://www.gaussianwaves.com/2014/07/power-delay-profile/)
[From Channel Impulse Response (CIR) to Power Delay Profile (PDP)](https://www.mathworks.com/matlabcentral/answers/44530-from-channel-impulse-response-cir-to-power-delay-profile-pdp)
"""
csi = csidata.csi
pdp = np.power(np.abs(phy_ifft(csi, scidx(20, 2), axis=1)),
2.0) / csi.shape[1]
norm = lambda x: x / np.max(x)
plt.figure()
plt.plot(norm(pdp[index, :12].reshape(12, -1)))
plt.title('Anechoic Chamber: $N_{SS}=%d$' % (sum(csi[index, 0, 0] != 0)))
plt.xlabel('sample index')
plt.ylabel('PDP')
plt.tight_layout()
plt.show()
def derotate_formual16(csi, phi, bw=20, ng=2):
"""De-rotate CSI phase using (16)"""
k = scidx(bw, ng)[:, np.newaxis, np.newaxis]
phi = phi[:, np.newaxis, np.newaxis, np.newaxis]
return np.exp(1j * phi * k) * csi
def derotate_linear(csi, bw=20, ng=2):
"""Windowing/CP removal effect compensation"""
k = scidx(bw, ng)[:, np.newaxis, np.newaxis]
phi = -7e-5 * k**3 + 3e-5 * k**2 + 0.05 * k
return np.exp(-1j * phi) * csi
def formula_16(csi):
"""formula_16"""
k = scidx(20, 2)[:, np.newaxis, np.newaxis]
phi = formula_15(csi)[:, np.newaxis, np.newaxis, np.newaxis]
return np.exp(1j * phi * k) * csi