Fix target function for subsampling alignment

2 files changed, 15 insertions, 2 deletions

diff --git a/dpd/src/Measure.py b/dpd/src/Measure.py
index 0f90a44..ba71917 100644
--- a/dpd/src/Measure.py
+++ b/dpd/src/Measure.py
@@ -130,11 +130,24 @@ class Measure:
             tx_rx_frame_path = ('/tmp/tx_rx_sync1_' +
                                 datetime.datetime.now().isoformat() +
                                 '.pdf')
+            plt.subplot(311)
             plt.plot(np.abs(rxframe_aligned[:128]), label="rxframe")
             plt.plot(np.abs(txframe_aligned[:128]), label="txframe")
             plt.xlabel("Samples")
+            plt.ylabel("Absolute Value")
+            plt.legend()
+            plt.subplot(312)
+            plt.plot(np.real(rxframe_aligned[:128]), label="rxframe_real", alpha=0.5, linestyle=":")
+            plt.plot(np.real(txframe_aligned[:128]), label="txframe_real", alpha=0.5, linestyle=":")
+            plt.xlabel("Samples")
             plt.ylabel("Real Part")
             plt.legend()
+            plt.subplot(313)
+            plt.plot(np.imag(rxframe_aligned[:128]), label="rxframe_imag", alpha=0.5, linestyle="--")
+            plt.plot(np.imag(txframe_aligned[:128]), label="txframe_imag", alpha=0.5, linestyle="--")
+            plt.xlabel("Samples")
+            plt.ylabel("Imaginary Part")
+            plt.legend()
             plt.savefig(tx_rx_frame_path)
             plt.clf()
 
diff --git a/dpd/src/subsample_align.py b/dpd/src/subsample_align.py
index eda1dce..4dc854b 100755
--- a/dpd/src/subsample_align.py
+++ b/dpd/src/subsample_align.py
@@ -52,7 +52,7 @@ def subsample_align(sig, ref_sig):
 
         corr_sig = np.fft.ifft(rotate_vec * fft_sig)
 
-        return -np.abs(np.sum(corr_sig * ref_sig))
+        return -np.abs(np.sum(corr_sig.conjugate() * ref_sig))
 
     optim_result = optimize.minimize_scalar(correlate_for_delay, bounds=(-1,1), method='bounded', options={'disp': True})
 
@@ -61,7 +61,7 @@ def subsample_align(sig, ref_sig):
 
         #print("Found subsample delay = {}".format(best_tau))
 
-        if 0:
+        if 1:
             corr = np.vectorize(correlate_for_delay)
             ixs = np.linspace(-1, 1, 100)
             taus = corr(ixs)