task_data_size = task_transmission_data["task_data_size"]
task_size[task_id] += task_data_size
for i, size in enumerate(task_size):
task_need_size = task_list[i]["data_size"]
task_need_size_list[i] = task_need_size * 1024 * 1024 * 0.5
diff_size.append(task_need_size * 1024 * 1024 * 0.5 - size)
if size >= task_need_size * 1024 * 1024 * 0.5:
finished[i] = 1
"""
*****************************************************************************************************************************************
"""
potential_value_list = np.zeros(node_num_list[experiment_median_no])
for i in range(node_num_list[experiment_median_no]):
potential_value = compute_potential_value(task_id_under_each_node_list[i], finished)
potential_value_list[i] = potential_value
max_potential_value[i] = max(potential_value, max_potential_value[i])
"""
*****************************************************************************************************************************************
completed_ratio: 任务完成率 = 所有完成任务 / 车辆范围内所有任务
channel_utilization_efficiency: 信道利用率 = 传输数据量 / （信道数量 * 时间）
social_welfare: 社会福利 = 所有节点的势函数值之和
"""
completed_num = 0
for isFinish in finished:
if isFinish == 1:
completed_num += 1
completed_ratio = completed_num / len(union_task_id_set)
def ipus_get_evaluation(strategy_list, experiment_median_num, node_num):
"""
:argument
strategy_list: 每个节点选择策略组成的列表
experiment_median_num： 从第几个experiment_median文件中读取实验设置
node_num：对应experiment_median 实验中具有任务的节点数量
:return
completed_ratio: 任务完成率 = 所有完成任务 / 车辆范围内所有任务
channel_utilization_efficiency: 信道利用率 = 传输数据量 / （信道数量 * 时间）
social_welfare: 社会福利 = 所有节点的势函数值之和
"""
pickle_file = Path(
load_experiment_median_from_pickle(experiment_median_num))
fp = pickle_file.open("rb")
iteration = pickle.load(fp)
fixed_edge_node = pickle.load(fp)
edge_vehicle_node = pickle.load(fp)
fixed_distance_matrix = pickle.load(fp)
mobile_distance_matrix = pickle.load(fp)
task_list = pickle.load(fp)
node_num_not_used = pickle.load(fp)
fixed_node_num = pickle.load(fp)
mobile_node_num = pickle.load(fp)
max_potential_value = pickle.load(fp)
useful_channel_under_node = pickle.load(fp)
task_id_under_each_node_list = pickle.load(fp)
usable_channel_of_all_nodes = pickle.load(fp)
task_time_limitation_of_all_nodes = pickle.load(fp)
combination_and_strategy_length_of_all_nodes = pickle.load(fp)
union_task_id_set = set()
for task_id_under_each_node in task_id_under_each_node_list:
task_id_set = set(task_id_under_each_node)
union_task_id_set = union_task_id_set | task_id_set
white_gaussian_noise = settings.WHITE_GAUSSIAN_NOISE
antenna_constant = settings.ANTENNA_CONSTANT
path_loss_exponent = settings.PATH_LOSS_EXPONENT
sub_channel_bandwidth = settings.SUB_CHANNEL_BANDWIDTH
selected_strategy = dict()
selected_strategy_no = dict()
"""
————————————————————————————————————————————————————————————————————————————————————
选择策略
————————————————————————————————————————————————————————————————————————————————————
"""
for i in range(node_num):
combination_and_strategy_length = combination_and_strategy_length_of_all_nodes[
i]
decimal_num = strategy_list[i]
x_base = combination_and_strategy_length["length_of_combination"][0]
x_base_num = list()
if decimal_num < 0:
print("decimal_num < 0")
while True:
decimal_num, remainder = divmod(decimal_num, x_base)
x_base_num.append(remainder)
if decimal_num <= 0:
break
if len(x_base_num) < 10:
for i in range(10 - len(x_base_num)):
x_base_num.append(0)
x_base_num.reverse()
strategy = constructor_of_strategy(
x_base_num=x_base_num,
combination=combination_and_strategy_length[
"combination_of_task_and_time"])
print("strategy")
print(strategy)
selected_strategy[str(i)] = strategy
selected_strategy_no[str(i)] = strategy_list[i]
# 更新信道分配
print(selected_strategy)
for i in range(len(selected_strategy)):
print(selected_strategy[str(i)])
useful_channel_under_node[i] = update_useful_channel(
selected_strategy[str(i)], useful_channel_under_node[i])
task_transmission_data_dict_of_all_nodes = dict()
"""
________________________________________________________________________________________________________________________________________-
计算传输数据
*****************************************************************************************************************************************
"""
node_type = settings.NODE_TYPE_FIXED
for fixed_node_id in range(fixed_node_num):
# print_to_console(task_id_under_each_node_list)
task_id_list = task_id_under_each_node_list[fixed_node_id]
node_strategy = selected_strategy[str(fixed_node_id)]
task_transmission_data_list = []
for task_id in task_id_list:
task_data_size = 0
for node_strategy_no in range(len(node_strategy)): # 对于节点的每个信道上的分配
allocated_channel_no = useful_channel_under_node[
fixed_node_id]["node_channel"][
node_strategy_no] # 该行策略分配对应的信道编号
if task_id == node_strategy[node_strategy_no][0]: # 为这个任务分配了信道
task_time = node_strategy[node_strategy_no][1] # 信道的分配时长
signal_list = list() # 保存每个边缘节点使用相同信道时传输到任务上的信号
for node_no, useful_channel in enumerate(
useful_channel_under_node[:fixed_node_num]):
node_channel = useful_channel["node_channel"]
channel_status = useful_channel["channel_status"]
for i, channel_no in enumerate(node_channel):
if allocated_channel_no == channel_no:
if channel_status[i] > 0:
distance = fixed_distance_matrix[node_no][
task_id]
channel_fading_gain = random_channel_fading_gain(
)
transmission_power = fixed_edge_node[
node_no]["channel_power"]
signal_value = np.square(
channel_fading_gain
) * antenna_constant * np.power(
distance, 0 - path_loss_exponent
) * transmission_power
signal = {
"node_type": settings.NODE_TYPE_FIXED,
"node_id": node_no,
"signal": signal_value[0]
}
signal_list.append(signal)
for node_no, useful_channel in enumerate(
useful_channel_under_node[fixed_node_num:]):
node_channel = useful_channel["node_channel"]
channel_status = useful_channel["channel_status"]
for i, channel_no in enumerate(node_channel):
if allocated_channel_no == channel_no:
if channel_status[i] > 0:
distance = mobile_distance_matrix[node_no][
task_id]
channel_fading_gain = random_channel_fading_gain(
)
transmission_power = edge_vehicle_node[
node_no]["channel_power"]
signal_value = np.square(
channel_fading_gain
) * antenna_constant * np.power(
distance, 0 - path_loss_exponent
) * transmission_power
signal = {
"node_type":
settings.NODE_TYPE_MOBILE,
"node_id":
node_no + int(len(fixed_edge_node)),
"signal":
signal_value
}
signal_list.append(signal)
inter_signal_value = 0
interference = 0
for signal_dict in signal_list:
if signal_dict["node_type"] == node_type:
if signal_dict["node_id"] == fixed_node_id:
inter_signal_value = signal_dict["signal"]
else:
interference += signal_dict["signal"]
else:
interference += signal_dict["signal"]
SINR = inter_signal_value / (interference +
white_gaussian_noise)
channel_data_size = task_time * sub_channel_bandwidth * np.log2(
1 + SINR)
task_data_size += channel_data_size
task_transmission_data = {
"task_id": task_id,
"task_data_size": task_data_size
}
task_transmission_data_list.append(task_transmission_data)
task_transmission_data_dict_of_all_nodes[str(
fixed_node_id)] = task_transmission_data_list
node_type = settings.NODE_TYPE_MOBILE
for mobile_node_id in range(fixed_node_num, node_num):
task_id_list = task_id_under_each_node_list[mobile_node_id]
node_strategy = selected_strategy[str(mobile_node_id)]
task_transmission_data_list = []
for task_id in task_id_list:
task_data_size = 0
for node_strategy_no in range(len(node_strategy)):
allocated_channel_no = useful_channel_under_node[
mobile_node_id]["node_channel"][
node_strategy_no] # 该行策略分配对应的信道编号
if task_id == node_strategy[node_strategy_no][0]:
task_time = node_strategy[node_strategy_no][1]
signal_list = list()
for node_no, useful_channel in enumerate(
useful_channel_under_node[:fixed_node_num]):
node_channel = useful_channel["node_channel"]
channel_status = useful_channel["channel_status"]
for i, channel_no in enumerate(node_channel):
if allocated_channel_no == channel_no:
if channel_status[i] > 0:
distance = fixed_distance_matrix[node_no][
task_id]
channel_fading_gain = random_channel_fading_gain(
)
transmission_power = fixed_edge_node[
node_no]["channel_power"]
signal_value = np.square(
channel_fading_gain
) * antenna_constant * np.power(
distance, 0 - path_loss_exponent
) * transmission_power
signal = {
"node_type": settings.NODE_TYPE_FIXED,
"node_id": node_no,
"signal": signal_value
}
signal_list.append(signal)
for node_no, useful_channel in enumerate(
useful_channel_under_node[fixed_node_num:]):
node_channel = useful_channel["node_channel"]
channel_status = useful_channel["channel_status"]
for i, channel_no in enumerate(node_channel):
if allocated_channel_no == channel_no:
if channel_status[i] > 0:
distance = mobile_distance_matrix[node_no][
task_id]
channel_fading_gain = random_channel_fading_gain(
)
transmission_power = edge_vehicle_node[
node_no]["channel_power"]
signal_value = np.square(
channel_fading_gain
) * antenna_constant * np.power(
distance, 0 - path_loss_exponent
) * transmission_power
signal = {
"node_type":
settings.NODE_TYPE_MOBILE,
"node_id":
node_no + int(len(fixed_edge_node)),
"signal":
signal_value
}
signal_list.append(signal)
inter_signal_value = 0
interference = 0
for signal_dict in signal_list:
if signal_dict["node_type"] == node_type:
if signal_dict["node_id"] == mobile_node_id:
inter_signal_value = signal_dict["signal"]
else:
interference += signal_dict["signal"]
else:
interference += signal_dict["signal"]
SINR = inter_signal_value / (interference +
white_gaussian_noise)
# print(SINR)
channel_data_size = task_time * sub_channel_bandwidth * np.log2(
1 + SINR)
task_data_size += channel_data_size
task_transmission_data = {
"task_id": task_id,
"task_data_size": task_data_size
}
task_transmission_data_list.append(task_transmission_data)
task_transmission_data_dict_of_all_nodes[str(
mobile_node_id)] = task_transmission_data_list
"""
*****************************************************************************************************************************************
计算 任务完成数
*****************************************************************************************************************************************
"""
finished = np.zeros(len(task_list))
task_size = np.zeros(len(task_list))
task_need_size_list = np.zeros(len(task_list))
diff_size = []
for key in task_transmission_data_dict_of_all_nodes.keys():
task_transmission_data_list = task_transmission_data_dict_of_all_nodes[
key]
for task_transmission_data in task_transmission_data_list:
task_id = task_transmission_data["task_id"]
task_data_size = task_transmission_data["task_data_size"]
task_size[task_id] += task_data_size
for i, size in enumerate(task_size):
task_need_size = task_list[i]["data_size"]
task_need_size_list[i] = task_need_size * 1024 * 1024 * 0.5
diff_size.append(task_need_size * 1024 * 1024 * 0.5 - size)
if size >= task_need_size * 1024 * 1024 * 0.5:
finished[i] = 1
"""
*****************************************************************************************************************************************
"""
potential_value_list = np.zeros(node_num)
for i in range(node_num):
potential_value = compute_potential_value(
task_id_under_each_node_list[i], finished)
potential_value_list[i] = potential_value
max_potential_value[i] = max(potential_value, max_potential_value[i])
"""
*****************************************************************************************************************************************
completed_ratio: 任务完成率 = 所有完成任务 / 车辆范围内所有任务
channel_utilization_efficiency: 信道利用率 = 传输数据量 / （信道数量 * 时间）
social_welfare: 社会福利 = 所有节点的势函数值之和
"""
completed_num = 0
print("finished")
print(finished)
for isFinish in finished:
if isFinish == 1:
completed_num += 1
completed_ratio = completed_num / len(union_task_id_set)
channel_utilization_num = 0
for strategy in strategy_list:
for strategy_channel in strategy:
if strategy_channel[0] != -1:
channel_time = strategy_channel[1]
channel_utilization_num += channel_time
channel_utilization_efficiency = sum(task_size) / channel_utilization_num
social_welfare = sum(potential_value_list)
return {
"completed_ratio": completed_ratio,
"channel_utilization_efficiency": channel_utilization_efficiency,
"social_welfare": social_welfare
}