# y1=np.array(y1)
# print(y1.shape)

# Network Parameters
n_hidden_1 = 30  # 1st layer num features
n_hidden_2 = 20  # 2nd layer num features
n_hidden_3 = 10
n_hidden_4 = 5
n_input = 90

if os.path.isdir(os.getcwd() + '/model'):
    shutil.rmtree(os.getcwd() + '/model')
os.mkdir(os.getcwd() + '/model')
for test_i in range(1, 18):  #we must create a sesession for each of loc
    os.mkdir(os.getcwd() + '/model/Test' + str(test_i))
    csi = np.squeeze(np.array(get_csi(test_i)))
    # create and train a graph for each point
    for k in range(16):  # for each test we have 16 train point in sn2
        os.mkdir(os.getcwd() + '/model/Test' + str(test_i) + '/' + str(k + 1))
        # tf Graph input (only pictures)
        X = tf.placeholder("float", [None, n_input])

        weights = {
            'encoder_h1':
            tf.Variable(tf.random_normal([n_input, n_hidden_1]), name='w1'),
            'encoder_h2':
            tf.Variable(tf.random_normal([n_hidden_1, n_hidden_2]), name='w2'),
            'encoder_h3':
            tf.Variable(tf.random_normal([n_hidden_2, n_hidden_3]), name='w3'),
            'encoder_h4':
            tf.Variable(tf.random_normal([n_hidden_3, n_hidden_4]), name='w4'),
Пример #2
0
import numpy as np

from get_v3 import get_csi
import os
import shutil

if os.path.isdir(os.getcwd() + '/Models'):

    shutil.rmtree(os.getcwd() + '/Models')

os.mkdir(os.getcwd() + '/Models')

random_points = list(
    np.array([1, 2, 4, 6, 7, 8, 10, 12, 13, 14, 16, 17, 19]) - 1)
for k in range(1, 14):
    csi = get_csi(k, random_points)
    csi = np.squeeze(csi)
    print(csi.shape)

    point1 = csi[0:20]
    point2 = csi[20:40]
    point3 = csi[40:60]
    point4 = csi[60:80]
    point5 = csi[80:100]
    point6 = csi[100:120]
    point7 = csi[120:140]
    point8 = csi[140:160]
    point9 = csi[160:180]
    point10 = csi[180:200]
    point11 = csi[200:220]
    point12 = csi[220:240]
import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt
#from read_from_db import read_csi_from_db
from get_v3 import get_csi
import os
import shutil

if os.path.isdir(os.getcwd()+'/Models'):

        shutil.rmtree(os.getcwd()+'/Models')
        #print("1")
os.mkdir(os.getcwd()+'/Models')

for k in range(1,20):
    csi = get_csi(k) #360*1*90
    csi = np.squeeze(csi) #360*90   20 packet from each of 18 Location -- 
    point1=csi[0:20]
    point2=csi[20:40]
    point3=csi[40:60]
    point4=csi[60:80]
    point5=csi[80:100]
    point6=csi[100:120]
    point7=csi[120:140]
    point8=csi[140:160]
    point9=csi[160:180]
    point10=csi[180:200]
    point11=csi[200:220]
    point12=csi[220:240]
    point13=csi[240:260]
    point14=csi[260:280]
# Network Parameters
n_hidden_1 = 30  # 1st layer num features
n_hidden_2 = 20  # 2nd layer num features
n_hidden_3 = 10
n_hidden_4 = 5
n_input = 90

if os.path.isdir(os.getcwd() + '/model'):
    shutil.rmtree(os.getcwd() + '/model')
os.mkdir(os.getcwd() + '/model')
for test_i in range(1, 17):  #we must create a session for each of loc
    random_point = list(
        np.array([1, 2, 4, 5, 6, 7, 9, 10, 11, 12, 13, 15, 16, 17, 18, 19]) -
        1)
    os.mkdir(os.getcwd() + '/model/Test' + str(test_i))
    csi = np.squeeze(np.array(get_csi(random_point[test_i - 1], random_point)))
    # create and train a graph for each point

    random_point.pop(test_i - 1)
    for k in range(len(random_point)):
        os.mkdir(os.getcwd() + '/model/Test' + str(test_i) + '/' + str(k + 1))
        # tf Graph input (only pictures)
        X = tf.placeholder("float", [None, n_input])

        weights = {
            'encoder_h1':
            tf.Variable(tf.random_normal([n_input, n_hidden_1]), name='w1'),
            'encoder_h2':
            tf.Variable(tf.random_normal([n_hidden_1, n_hidden_2]), name='w2'),
            'encoder_h3':
            tf.Variable(tf.random_normal([n_hidden_2, n_hidden_3]), name='w3'),
import tensorflow as tf
import numpy as np
# import matplotlib.pyplot as plt
#from read_from_db import read_csi_from_db
from get_v3 import get_csi
import os
import shutil

if os.path.isdir(os.getcwd()+'/Models'):

        shutil.rmtree(os.getcwd()+'/Models')
        #print("1")
os.mkdir(os.getcwd()+'/Models')

for k in range(1,18):
    csi = get_csi(k)
    csi = np.squeeze(csi) # 320*90  --> [20*16,90]
    point1=csi[0:20]
    point2=csi[20:40]
    point3=csi[40:60]
    point4=csi[60:80]
    point5=csi[80:100]
    point6=csi[100:120]
    point7=csi[120:140]
    point8=csi[140:160]
    point9=csi[160:180]
    point10=csi[180:200]
    point11=csi[200:220]
    point12=csi[220:240]
    point13=csi[240:260]
    point14=csi[260:280]

#num_of_SPs=[6,10,14,17]
num_of_SPs=17

for sp in [num_of_SPs]:
    total_elapsed_time=[]
    total_memory_used=[]
    total_memory_perc=[]
    random_list=random.sample(range(17), sp)
    if os.path.isdir(os.getcwd()+r'\Models'):
            shutil.rmtree(os.getcwd()+r'\Models')
    os.mkdir(os.getcwd()+r'\Models')

    for test_i in [random_list.index(random_list[-1])]:  # we must create a session for each of loc --> model all other sps with one net!!
        csi = get_csi(random_list[test_i],random_list)
        csi = np.squeeze(csi)
        st = time.time()
        points=[]
        for t in range(sp):
            points.append(csi[t*20:(t+1)*20])

        total=points

        # Parameters
        learning_rate = 0.01
        training_epochs = 1000
        display_step = 50
        n_labels = sp-1
        #
for sp in [6]:
    total_elapsed_time = []
    total_memory_used = []
    total_memory_perc = []
    random_list = random.sample(range(19), sp)

    if os.path.isdir(os.getcwd() + r'\model'):
        shutil.rmtree(os.getcwd() + r'\model')
    os.mkdir(os.getcwd() + r'\model')
    os.mkdir(os.getcwd() + r'\model\-' + str(sp))
    for test_i in range(len(random_list) -
                        1):  #we must create a session for each of loc
        os.mkdir(os.getcwd() + r'\model\-' + str(sp) + r'\Test-' +
                 str(test_i + 1))
        csi = np.squeeze(np.array(get_csi(random_list[test_i], random_list)))
        # create and train a graph for each point
        st = time.time()
        # tf Graph input (only pictures)
        X = tf.placeholder("float", [None, n_input])

        weights = {
            'encoder_h1':
            tf.Variable(tf.random_normal([n_input, n_hidden_1]), name='w1'),
            'encoder_h2':
            tf.Variable(tf.random_normal([n_hidden_1, n_hidden_2]), name='w2'),
            'encoder_h3':
            tf.Variable(tf.random_normal([n_hidden_2, n_hidden_3]), name='w3'),
            'encoder_h4':
            tf.Variable(tf.random_normal([n_hidden_3, n_hidden_4]), name='w4'),
            'decoder_h1':