-
Notifications
You must be signed in to change notification settings - Fork 5
/
user.py
219 lines (181 loc) · 10.4 KB
/
user.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
import numpy as np
import matplotlib.pyplot as plt
import os
from utils import * # Load utility functions
from system_paras import * # Load system parameters
rng = np.random.default_rng()
# Calculate the next position of user i, (r2,varphi2), based on the previous location,
# (r1, varphi1), the movement direction (phi), and the freedom of movement (stddev_delta_varphi)
def update_location(phi, r1, varphi1, stddev_delta_varphi, mean_delta_r):
stddev_distance = mean_delta_r/3
delta_r = mean_delta_r + stddev_distance * rng.standard_normal()
delta_r = np.where(delta_r > 0, delta_r, 0)
# delta_varphi = mean_delta_varphi *np.pi + stddev_delta_varphi * rng.standard_normal() # normally distributed with mean=0 and variance=(stddev_delta_varphi)**2
delta_varphi = (phi-varphi1) + stddev_delta_varphi * rng.standard_normal() # normally distributed with mean=(phi-varphi1) and variance=(stddev_delta_varphi)**2
if delta_varphi > np.pi: # normalize so that delta_varphi in range [0,2pi]
delta_varphi -= 2*np.pi
elif delta_varphi < -np.pi:
delta_varphi += 2*np.pi
r2 = np.sqrt( r1**2 + delta_r**2 + 2*r1*delta_r*np.cos(delta_varphi) )
theta = np.arccos( (r1**2+r2**2-delta_r**2)/(2*r1*r2) ) # angle between vectors r1 and delta_r, could be negative (i.e., a clockwise direction)
varphi2 = np.where( delta_varphi>0, varphi1+theta, varphi1-theta )
if varphi2 > np.pi: # normalize so that varphi in range [0,2pi]
varphi2 -= 2*np.pi
elif varphi2 < np.pi:
varphi2 += 2*np.pi
return (r2,varphi2)
class User:
def __init__(self, phi, r0, stddev_delta_varphi, varphi0, mean_delta_r):
self.arrival_task = np.zeros(time_max)
self.qlen_thres = Mbits(1000) # threshold for long-term average of qlen (qlen < qlen_thres), initiated as infinite
# Movement: location = (radius, varphi) in the cylinderic coordinate
self.phi = phi # movement direction
self.radius = np.zeros(time_max)
self.varphi = np.zeros(time_max)
self.varphi[0] = varphi0 # initial varphi
self.radius[0] = r0 # initial radius
self.mean_delta_r = mean_delta_r
for t in range(1, time_max):
if isUserLocationFixed == False:
(r2,varphi2) = update_location(phi=self.phi, r1=self.radius[t-1], \
varphi1=self.varphi[t-1], stddev_delta_varphi=stddev_delta_varphi,
mean_delta_r=self.mean_delta_r)
self.radius[t] = r2
self.varphi[t] = varphi2
else:
self.radius[t] = self.radius[t-1]
self.varphi[t] = self.varphi[t-1]
# Channel gain of the user-UAV link
theta = np.arctan(uav_altitude / self.radius) # elevation angle btw user and uav
PLOS = 1/( 1 + a_LOS*np.exp( - b_LOS * ( theta - a_LOS ) ) ) # size = (time_max,)
fading = dB( (mean_fading_log + stddev_fading_log * rng.standard_normal(time_max)) )
self.channel_gain_nofading = ( PLOS + zeta_LOS*(1-PLOS) ) * g0 \
/ ( uav_altitude**2 + self.radius**2 )**(gamma/2)
self.channel_gain = self.channel_gain_nofading * fading
# Channel gain of the user-mBS link
fading_BS = dB( (mean_fading_log_BS + stddev_fading_log_BS * rng.standard_normal(time_max)) )
self.channel_gain_BS = fading_BS * g0 / d_macroBS**gamma
self.cpu_frequency = np.zeros(time_max)
self.tasks_computed_locally = np.zeros(time_max)
self.tasks_offloaded_to_server = np.zeros(time_max)
self.power_local_computation = np.zeros(time_max)
self.power_transmit = np.zeros(time_max)
self.pw_total = np.zeros(time_max)
self.queue_length = np.zeros(time_max)
# self.queue_length[0] += 1 # at time = 0, 1 bit in queue
self.vq_qlen_penalty = np.zeros(time_max)
# self.vq_qlen_penalty[0] += 1 # at time = 0, 1 bit in queue
##### Compute tasks locally:
def opt_fcpu_local(self, t, tasks_backlog, VQ_local_i):
'''
Arguments:
- tasks_backlog = tasks (in the queue at the end of the previous slot) - tasks (offloaded)
- VQ_local = virtual queue for penalty if qlen > a specific threshold
Return:
cpu_frequency, tasks_computed_locally, power_local_computation
'''
tasks_backlog_sum = tasks_backlog + VQ_local_i
cpu_freq_optimal = np.min([ np.min([fcpu_max, tasks_backlog * cycles_per_bit / slot_len]),
np.sqrt(tasks_backlog_sum*slot_len/(3*kappa*Vlyapunov*cycles_per_bit)) ]) # select the optimal cpu frequency
pw_computation_local = kappa*cpu_freq_optimal**3 # power consumption for local computation
tasks_computed_locally = slot_len*cpu_freq_optimal/cycles_per_bit # update task computed locally
return (cpu_freq_optimal, pw_computation_local, tasks_computed_locally)
##### Update the queue length after task computation and task offloading
def update_queue(self, t):
if t+1 < time_max: # so that (t+1) <= (time_max-1)
self.queue_length[t+1] = self.arrival_task[t] + np.max([0, self.queue_length[t] - \
(self.tasks_computed_locally[t] + self.tasks_offloaded_to_server[t])])
self.vq_qlen_penalty[t+1] = np.max([0, self.vq_qlen_penalty[t] + scale_vq * (self.queue_length[t+1] - self.qlen_thres)])
##### Update power consumption, note: power, not energy -> do not count slot_len
def update_power(self, t):
self.pw_total[t] = self.power_local_computation[t] + self.power_transmit[t]
def gen_users():
pickle_fn = "users (time={t1}s, slot={t2:.2}s).pickle".format(t1=total_time, t2=slot_len, A=Amean/1e6)
chgain_fn = "users-channel-gain (time={t1}s, slot={t2:.2}s).png".format(t1=total_time, t2=slot_len)
locations_fn = "users-location (time={t1}s, slot={t2:.2}s).png".format(t1=total_time, t2=slot_len)
if os.path.exists(os.path.join(users_folder, pickle_fn))==True:
import warnings
warnings.warn(f'Data of users existed, filepath = "{os.path.join(users_filepath, pickle_fn)}"')
else:
users = [] # list of users
list_of_users = [] # list of users' properties
for i in range(num_users):
r0 = rng.integers(50,150) # initial radical distance
varphi0 = rng.random()*2*np.pi # initial angular coordinate
phi = rng.random()*2*np.pi # movement direction
mean_delta_r0 = mean_velocity*slot_len
list_of_users.append( (r0, varphi0, phi, mean_delta_r0) )
# generate users
for idx, (r0, varphi0, phi, mean_delta_r) in enumerate(list_of_users):
users.append( User(phi=phi, r0=r0,
stddev_delta_varphi=stddev_delta_varphi,
varphi0=varphi0,
mean_delta_r=mean_delta_r
) )
# Save data to a pickle file
filepath = os.path.join(users_folder, pickle_fn)
save_data(users, filepath)
print('Generated users successfully, filepath="{}"'.format(filepath))
# For plotting figures
lines_color = ['-b', '-g', '-r', '-c', '-m', '-k']
lines_color_nofading = ['.b', '.g', '.r', '.c', '.m', '.k']
dots_color = ['ob', 'og', 'or', 'oc', 'om', 'ok']
n_plot = 3 # must be strictly less than len(lines_color)
##### Test 1 : plot radical distance ri(t) versus time
# fig1 = plt.figure()
# for idx, user in enumerate(users):
# plt.plot(range(0,time_max),user.radius,lines_color[idx], label=f"users[{idx}]")
# plt.xlabel('Time')
# plt.ylabel('Radical distance, r(t)')
# plt.grid(True)
# # fig1.show()
# plt.savefig('radical distance.png')
##### Test 2 : plot real-time locations of users on the ground
plt.figure() # create a figure
for idx, user in enumerate(users):
if idx >= n_plot:
break
x = user.radius*np.cos(user.varphi)
y = user.radius*np.sin(user.varphi)
plt.plot(x, y, lines_color[idx], label=f"users[{idx}]")
plt.plot(x[0], y[0], dots_color[idx])
plt.plot(0, 0, 'ok',label=f"UAV")
plt.legend()
plt.grid(True)
plt.xlabel('x (m)')
plt.ylabel('y (m)')
plt.savefig(os.path.join(users_folder, locations_fn))
##### Test 3 : Plotting radical distance, r_i(t), and locations of users, (x_i,y_i), wrt time
# fig, (ax1,ax2) = plt.subplots(1,2)
# for idx, user in enumerate(users):
# ax1.plot(range(0,time_max),user.radius,lines_color[idx], label=f"users[{idx}]")
# ax1.grid(True)
# ax1.set_xlabel('Time')
# ax1.set_ylabel('Radical distance, r(t)')
# for idx, user in enumerate(users):
# x = user.radius*np.cos(user.varphi)
# y = user.radius*np.sin(user.varphi)
# # plt.plot(range(0,time_max),users[0].radius,'-r')
# ax2.plot(x, y, lines_color[idx], label=f"users[{idx}]")
# ax2.plot(x[0], y[0], dots_color[idx])
# ax2.grid(True)
# ax2.set_xlabel('x (m)')
# ax2.set_ylabel('y (m)')
# fig.show()
# fig.savefig(dir_name + f'/location_merge.png')
##### Test 4 : plotting channel gain, h_i(t)
plt.figure()
tmax = -1
for idx, user in enumerate(users):
if idx >= n_plot:
break
plt.plot(range(0,time_max)[:tmax], to_dB(user.channel_gain[:tmax]), lines_color[idx], label=f"users[{idx}], h-UAV", linewidth=0.5)
# plt.plot(range(0,time_max)[:tmax], to_dB(user.channel_gain_nofading[:tmax]), lines_color_nofading[idx], markersize=2)
plt.plot(range(0,time_max)[:tmax], to_dB(user.channel_gain_BS[:tmax]), lines_color[idx], markersize=2)
plt.xlabel("Time")
plt.ylabel('Channel gain, h(t)')
plt.grid(True)
plt.legend()
plt.savefig(os.path.join(users_folder, chgain_fn))
if __name__ == "__main__":
gen_users()