#!/usr/bin/env python
# coding: utf-8

# ## Experiment 7 - Scipy's correlate vs. norm correlate

# The aim of this experiment is to determine which correlate is better: scipy's correlate or norm correlate

# In[1]:


import os, sys
nb_dir = os.path.split(os.getcwd())[0]
if nb_dir not in sys.path:
    sys.path.append(nb_dir)
    
get_ipython().run_line_magic('matplotlib', 'inline')
get_ipython().run_line_magic('load_ext', 'autoreload')
get_ipython().run_line_magic('autoreload', '2')


# In[3]:


from directdemod import source, chunker, comm, constants, filters, fmDemod, sink, amDemod, noaa, log
import numpy as np
import sys, getopt, logging
import matplotlib.pyplot as plt
log.log(console = False)


# To get syncs from a file we do,

# In[4]:


sigsrc = source.IQwav("D:\\noaa\\station1_yagi_SDRSharp_20170312_060959Z_137650kHz_IQ.wav")
noaaObj = noaa.noaa(sigsrc, constants.IQ_FREQOFFSET)
syncs = noaaObj.getAccurateSync


# In[11]:


syncA, syncB = syncs
print("SyncA:",syncA)
print("SyncB:",syncB)
print()
print("diffSyncA:",np.diff(syncA))
print("diffSyncB:",np.diff(syncB))
print()
print("abs(diffSyncA-1024000):", np.abs(np.diff(syncA) - 1024000))
print("abs(diffSyncB-1024000):", np.abs(np.diff(syncB) - 1024000))
print()
print("sum(abs(diffSyncA-1024000)):", np.sum(np.abs(np.diff(syncA) - 1024000)))
print("sum(abs(diffSyncB-1024000)):", np.sum(np.abs(np.diff(syncB) - 1024000)))


# ## Method
# 
# Experimental environment will be:
# same file and same everything except (correlate function, filter before correlation, needle type)

# # EXP A
# 2 runs: variables [1st with norm correlate, 2nd with scipys correlate]
# 
# constants [hamming filter of width 492, needle varies from 11 to 244]

# In[7]:


sigsrc = source.IQwav("D:\\noaa\\station1_yagi_SDRSharp_20170312_060959Z_137650kHz_IQ.wav")
noaaObj = noaa.noaa(sigsrc, constants.IQ_FREQOFFSET)
syncs = noaaObj.getAccurateSync
syncA1, syncB1 = syncs


# In[9]:


sigsrc = source.IQwav("D:\\noaa\\station1_yagi_SDRSharp_20170312_060959Z_137650kHz_IQ.wav")
noaaObj = noaa.noaa(sigsrc, constants.IQ_FREQOFFSET)
syncs = noaaObj.getAccurateSync
syncA2, syncB2 = syncs


# In[10]:


print("SyncA1:",syncA1)
print("SyncB1:",syncB1)
print()
print("diffSyncA1:",np.diff(syncA1))
print("diffSyncB1:",np.diff(syncB1))
print()
print("abs(diffSyncA1-1024000):", np.abs(np.diff(syncA1) - 1024000))
print("abs(diffSyncB1-1024000):", np.abs(np.diff(syncB1) - 1024000))
print()
print("sum(abs(diffSyncA1-1024000)):", np.sum(np.abs(np.diff(syncA1) - 1024000)))
print("sum(abs(diffSyncB1-1024000)):", np.sum(np.abs(np.diff(syncB1) - 1024000)))
print()
print("SyncA2:",syncA2)
print("SyncB2:",syncB2)
print()
print("diffSyncA2:",np.diff(syncA2))
print("diffSyncB2:",np.diff(syncB2))
print()
print("abs(diffSyncA2-1024000):", np.abs(np.diff(syncA2) - 1024000))
print("abs(diffSyncB2-1024000):", np.abs(np.diff(syncB2) - 1024000))
print()
print("sum(abs(diffSyncA2-1024000)):", np.sum(np.abs(np.diff(syncA2) - 1024000)))
print("sum(abs(diffSyncB2-1024000)):", np.sum(np.abs(np.diff(syncB2) - 1024000)))
print()
print("diffSyncA1-diffSyncA2:", [a-b for a,b in zip(np.diff(syncA1),np.diff(syncA2))])
print("diffSyncB1-diffSyncB2:", [a-b for a,b in zip(np.diff(syncB1),np.diff(syncB2))])
print()
print("IMP RESULTS")
print()
print("NORM CORR:")
print("\tMax Deviation in diffSyncA1", np.max(np.diff(syncA1))-np.min(np.diff(syncA1)))
print("\tMax Deviation in diffSyncB1", np.max(np.diff(syncB1))-np.min(np.diff(syncB1)))
print("\tCumulative deviation from ideal SyncA1", np.sum(np.abs(np.diff(syncA1) - 1024000)))
print("\tCumulative deviation from ideal SyncB1", np.sum(np.abs(np.diff(syncB1) - 1024000)))
print()
print("SCIPY:")
print("\tMax Deviation in diffSyncA2", np.max(np.diff(syncA2))-np.min(np.diff(syncA2)))
print("\tMax Deviation in diffSyncB2", np.max(np.diff(syncB2))-np.min(np.diff(syncB2)))
print("\tCumulative deviation from ideal SyncA2", np.sum(np.abs(np.diff(syncA2) - 1024000)))
print("\tCumulative deviation from ideal SyncB2", np.sum(np.abs(np.diff(syncB2) - 1024000)))


# ### Result: Norm Corr wins
# - min deviation from ideal
# - min deviation
# - Scipys correlate doesn't seem to work with this needle

# # EXP B
# 2 runs: variables [1st with norm correlate, 2nd with scipys correlate]
# 
# constants [hamming filter of width 492, needle varies from -0.5 to 0.5]

# In[13]:


sigsrc = source.IQwav("D:\\noaa\\station1_yagi_SDRSharp_20170312_060959Z_137650kHz_IQ.wav")
noaaObj = noaa.noaa(sigsrc, constants.IQ_FREQOFFSET)
syncs = noaaObj.getAccurateSync
syncA1, syncB1 = syncs


# In[12]:


sigsrc = source.IQwav("D:\\noaa\\station1_yagi_SDRSharp_20170312_060959Z_137650kHz_IQ.wav")
noaaObj = noaa.noaa(sigsrc, constants.IQ_FREQOFFSET)
syncs = noaaObj.getAccurateSync
syncA2, syncB2 = syncs


# In[14]:


print("SyncA1:",syncA1)
print("SyncB1:",syncB1)
print()
print("diffSyncA1:",np.diff(syncA1))
print("diffSyncB1:",np.diff(syncB1))
print()
print("abs(diffSyncA1-1024000):", np.abs(np.diff(syncA1) - 1024000))
print("abs(diffSyncB1-1024000):", np.abs(np.diff(syncB1) - 1024000))
print()
print("sum(abs(diffSyncA1-1024000)):", np.sum(np.abs(np.diff(syncA1) - 1024000)))
print("sum(abs(diffSyncB1-1024000)):", np.sum(np.abs(np.diff(syncB1) - 1024000)))
print()
print("SyncA2:",syncA2)
print("SyncB2:",syncB2)
print()
print("diffSyncA2:",np.diff(syncA2))
print("diffSyncB2:",np.diff(syncB2))
print()
print("abs(diffSyncA2-1024000):", np.abs(np.diff(syncA2) - 1024000))
print("abs(diffSyncB2-1024000):", np.abs(np.diff(syncB2) - 1024000))
print()
print("sum(abs(diffSyncA2-1024000)):", np.sum(np.abs(np.diff(syncA2) - 1024000)))
print("sum(abs(diffSyncB2-1024000)):", np.sum(np.abs(np.diff(syncB2) - 1024000)))
print()
print("diffSyncA1-diffSyncA2:", [a-b for a,b in zip(np.diff(syncA1),np.diff(syncA2))])
print("diffSyncB1-diffSyncB2:", [a-b for a,b in zip(np.diff(syncB1),np.diff(syncB2))])
print()
print("IMP RESULTS")
print()
print("NORM CORR:")
print("\tMax Deviation in diffSyncA1", np.max(np.diff(syncA1))-np.min(np.diff(syncA1)))
print("\tMax Deviation in diffSyncB1", np.max(np.diff(syncB1))-np.min(np.diff(syncB1)))
print("\tCumulative deviation from ideal SyncA1", np.sum(np.abs(np.diff(syncA1) - 1024000)))
print("\tCumulative deviation from ideal SyncB1", np.sum(np.abs(np.diff(syncB1) - 1024000)))
print()
print("SCIPY:")
print("\tMax Deviation in diffSyncA2", np.max(np.diff(syncA2))-np.min(np.diff(syncA2)))
print("\tMax Deviation in diffSyncB2", np.max(np.diff(syncB2))-np.min(np.diff(syncB2)))
print("\tCumulative deviation from ideal SyncA2", np.sum(np.abs(np.diff(syncA2) - 1024000)))
print("\tCumulative deviation from ideal SyncB2", np.sum(np.abs(np.diff(syncB2) - 1024000)))


# ### Result: Scipy wins
# - min deviation from ideal in scipy
# - min deviation in scipy

# # Final Result
# 
# Scipy's correlate got the best answer and it is much more faster than norm correlate. Here is the plot of the syncs extracted from whole signal

# In[16]:


sigsrc = source.IQwav("D:\\noaa\\station1_yagi_SDRSharp_20170312_060959Z_137650kHz_IQ.wav")
noaaObj = noaa.noaa(sigsrc, constants.IQ_FREQOFFSET)
syncs = noaaObj.getAccurateSync
syncA2, syncB2 = syncs


# In[18]:


plt.plot(np.diff(syncA2))
plt.plot(np.diff(syncB2))
plt.show()