def get_known_cdf_time(n_f):
time = np.zeros((lsnr.size, n_f.size), float)
for j, n in enumerate(n_f):
for i, c in enumerate(ccdf):
st = strategizer(c, x_cdf).summary(n)
t, p = strategy_time(st, c, x_cdf, n, False)
time[i, j] = t/p
return time
def get_arq_time(n_f, after_n):
time = np.zeros((lsnr.size, n_f.size), float)
for j, n in enumerate(n_f):
for i, c in enumerate(ccdf):
st = [after_n]
t, p = strategy_time(st, c, x_cdf, n, False)
time[i, j] = t/p
return time
def get_cdf_trace(Fs, t, lsnr_trace, n_f, length, known_cdfs, alpha):
out_throughput = np.zeros(10000)
out_time = np.zeros(10000)
if not known_cdfs:
learner = gaussian_ccdf_learner(x_cdf, alpha)
time = 0.
i = 0
while time < t[-1]:
# get the current ccdf
current_lsnr = lsnr_trace[np.nonzero(t>=time)[0][0]]
idx = np.clip(current_lsnr-np.amin(lsnr), 0, lsnr.size-1)
c = ccdf[int(idx)]
if int(idx)+1 < lsnr.size:
a = idx - int(idx)
c = c + a * (ccdf[int(idx)+1] - c)
# pick a strategy
if known_cdfs:
st = strategizer(c, x_cdf).summary(n_f)
else:
st = strategizer(learner.getstate(), x_cdf).summary(n_f)
# follow the strategy for one transmission
if not known_cdfs:
r = np.random.random()
n_success = x_cdf[np.where(r > c)[0]]
n_success = n_success[0] if n_success.size else x_cdf[-1]
learner.learn(n_success)
time_, p_success_ = strategy_time(st, c, x_cdf, n_f, False)
# time is in half symbols
time += time_/Fs/2
throughput = Fs*length*p_success_/(time_/2)
i += 1
if i % 100 == 0:
replace_line('known_cdfs=%s alpha=%.2f t=%5.1f ms SNR=%5.2f dB throughput=%4.1f Mbps capacity=%4.1f Mbps' % (known_cdfs, alpha, 1000*time, current_lsnr, throughput*1e-6, 12*np.log2(1+10.**(.1*current_lsnr))))
if i >= out_throughput.size:
out_throughput = np.r_[out_throughput, np.zeros(10000)]
out_time = np.r_[out_time, np.zeros(10000)]
out_throughput[i] = throughput
out_time[i] = time
replace_line('')
return out_throughput[:i], out_time[:i]