plt.plot(bandwidth, cost_of_all, '^-', linewidth=0.4, label='all selection')
plt.plot(bandwidth, cost_of_local, '<-', linewidth=0.4, label='only local selection')
plt.plot(bandwidth, cost_of_mec, '>-', linewidth=0.4, label='only MEC selection')
plt.plot(bandwidth, cost_of_all_15ge, '<-', linewidth=0.2, label='all selection of 15 UEs')
plt.plot(bandwidth, cost_of_all_20ge, '<-', linewidth=0.2, label='all selection of 20 UEs')
plt.plot(bandwidth, cost_of_all_25ge, '<-', linewidth=0.2, label='all selection of 25 UEs')
plt.grid(True) #显示网格
plt.xlabel('The Bandwidth of Channel')
plt.ylabel('Sum Cost')
plt.legend(loc='upper right') #图例右上角
plt.show()
data = DTE("./picture/pic3/all") ## TLIU
print(cost_of_all)
data.write(cost_of_all)
data = DTE("./picture/pic3/mec") ## TLIU
print(cost_of_mec)
data.write(cost_of_mec)
data = DTE("./picture/pic3/local") ## TLIU
print(cost_of_local)
data.write(cost_of_local)
data = DTE("./picture/pic3/all_15_UEs") ## TLIU
print(cost_of_all_15ge)
data.write(cost_of_all_15ge)
actions_set[iteration_actions[i]][0],
rff[i])
queue_relay_array[i].updateQx()
queue_relay_array[i].updateQy()
queue_relay_array[i].updateQz()
# reward step
reward_history.append(sum(reward))
for i in range(user_num):
wolf_agent_array[i].observe(reward=reward[i])
for i in range(user_num):
print(wolf_agent_array[i].pi_average)
# plt.plot(np.arange(len(reward_history)), reward_history, label="all")
# plt.title('wolf_dl2-dh6')
# # plt.show()
# data = DTE("./picture/pic1/wolf_dl2_dh6") ## TLIU
# print(OUTPUT)
# data.write(OUTPUT)
# plt.plot(np.arange(len(PR[1])), PR[1])
# plt.title('PR[1]')
# plt.show()
for i in range(user_num):
data = DTE("./picture/pic1/PR" + str(i)) ## TLIU
data.write(PR[i])
plt.plot(usernumber, cost_of_local, '<-', linewidth=0.4, label='only local selection')
plt.plot(usernumber, cost_of_mec, '>-', linewidth=0.4, label='only MEC selection')
plt.plot(usernumber, cost_of_all_6mhz, '<-', linewidth=0.2, label='all selection of 6mhz')
plt.plot(usernumber, cost_of_all_8mhz, '<-', linewidth=0.2, label='all selection of 8mhz')
plt.plot(usernumber, cost_of_all_12mhz, '<-', linewidth=0.2, label='all selection of 12mhz')
plt.grid(True) #显示网格
plt.xlabel('The number of UE')
plt.ylabel('Sum Cost')
plt.legend(loc='upper left') #图例右上角
plt.show()
data = DTE("./picture/pic2/all") ## TLIU
print(cost_of_all)
data.write(cost_of_all)
data = DTE("./picture/pic2/mec") ## TLIU
print(cost_of_mec)
data.write(cost_of_mec)
data = DTE("./picture/pic2/local") ## TLIU
print(cost_of_local)
data.write(cost_of_local)
data = DTE("./picture/pic2/all_6MHZ") ## TLIU
print(cost_of_all_6mhz)
data.write(cost_of_all_6mhz)
reward, bn, lumbda, rff = game.step(actions=iteration_actions)
print("episode", episode, "reward", sum(reward))
OUTPUT.append(sum(reward))
for i in range(user_num):
#wolf agent act
# update_Queue_relay
queue_relay_array[i].lumbda = lumbda[i]
queue_relay_array[i].updateQ(bn[i],
actions_set[iteration_actions[i]][0],
rff[i])
queue_relay_array[i].updateQx()
queue_relay_array[i].updateQy()
queue_relay_array[i].updateQz()
# reward step
reward_history.append(sum(reward))
for i in range(user_num):
wolf_agent_array[i].observe(reward=reward[i])
for i in range(user_num):
print(wolf_agent_array[i].pi_average)
plt.plot(np.arange(len(reward_history)), reward_history, label="all")
plt.show()
data = DTE("./picture/pic2/wolf") ## TLIU
print(OUTPUT)
data.write(OUTPUT)
# data = DTE("./picture/pic1/wolf_dl2_dh6") ## TLIU
# print(OUTPUT)
# data.write(OUTPUT)
if __name__ == '__main__':
dl = [0.0005, 0.001, 0.0015, 0.002, 0.0025, 0.003, 0.0035, 0.004]
dh = [0.004, 0.006, 0.008, 0.01, 0.012, 0.14, 0.016]
reward_dif_wolf = []
# reward_dif_wolf.append(difwolfphc.wolf_cal_reward(DL=0.001, DH=0.006))
# print('reward_differ_wolf',reward_dif_wolf)
# for i in range(len(dl)):
# for j in range(len(dh)):
i = 6
j = 5
if 1 == 1:
if 1 == 1:
# for j in range(len(dh)):
difwolfphc = differ_DL_DH()
print(
"==================================i,j ",
i, j)
reward_dif_wolf.append(
difwolfphc.wolf_cal_reward(DL=dl[i], DH=dh[j]))
data = DTE("./picture/pic1/differ_wolf_dl_dh") ## TLIU
print('reward_differ_wolf', reward_dif_wolf)
data.write(reward_dif_wolf)
OUTPUT.append(sum(reward))
for i in range(user_num):
#wolf agent act
# update_Queue_relay
queue_relay_array[i].lumbda = lumbda[i]
queue_relay_array[i].updateQ(bn[i], actions_set[iteration_actions[i]][0], rff[i])
queue_relay_array[i].updateQx()
queue_relay_array[i].updateQy()
queue_relay_array[i].updateQz()
# reward step
reward_history.append(sum(reward))
for i in range(user_num):
wolf_agent_array[i].observe(reward=reward[i])
for i in range(user_num):
print(wolf_agent_array[i].pi_average)
plt.plot(np.arange(len(reward_history)), reward_history, label="only local")
plt.show()
#
data = DTE("./picture/pic1/local-new") ## TLIU
print(OUTPUT)
data.write(OUTPUT)
#print('Q value :'+str(Q_array)+str(Qx_array)+str(Qy_array)+str(Qz_array))
reward, bn, lumbda, rff = game.step(actions=iteration_actions)
print("episode", episode, "reward", sum(reward))
OUTPUT.append(sum(reward))
for i in range(user_num):
#wolf agent act
# update_Queue_relay
queue_relay_array[i].lumbda = lumbda[i]
queue_relay_array[i].updateQ(bn[i],
actions_set[iteration_actions[i]][0],
rff[i])
queue_relay_array[i].updateQx()
queue_relay_array[i].updateQy()
queue_relay_array[i].updateQz()
# reward step
reward_history.append(sum(reward))
for i in range(user_num):
wolf_agent_array[i].observe(reward=reward[i])
for i in range(user_num):
print(wolf_agent_array[i].pi_average)
plt.plot(np.arange(len(reward_history)), reward_history, label="only mec")
plt.show()
data = DTE("./picture/pic1/mec") ## TLIU
print(OUTPUT)
data.write(OUTPUT)
reward, bn, lumbda, rff = game.step(actions=iteration_actions)
print("episode", episode, "reward", sum(reward))
OUTPUT.append(sum(reward))
for i in range(user_num):
# wolf agent act
# update_Queue_relay
queue_relay_array[i].lumbda = lumbda[i]
queue_relay_array[i].updateQ(bn[i],
actions_set[iteration_actions[i]][0],
rff[i])
queue_relay_array[i].updateQx()
queue_relay_array[i].updateQy()
queue_relay_array[i].updateQz()
# reward step
reward_history.append(sum(reward))
for i in range(user_num):
wolf_agent_array[i].observe(reward=reward[i])
for i in range(user_num):
print(wolf_agent_array[i].pi_average)
plt.plot(np.arange(len(reward_history)), reward_history, label="all")
plt.title('wolf_dl2-dh5')
plt.show()
data = DTE("./picture/pic1/wolf_dl2_dh5") ## TLIU
print(OUTPUT)
data.write(OUTPUT)