Fig. 1. Schematic of atmospheric turbulence channel simulated by phase screens

Fig. 2. Flow chart of SVM detection algorithm

Fig. 3. Normalized received intensity distributions when Rytov variance is (a) 0.1, (b) 0.3, (c) 0.5 and (d) 0.7, respectively

Fig. 4. Mean normalized intensities change with time when Rytov variance is (a) 0.1, (b) 0.3, (c) 0.5 and (d) 0.7, respectively

Fig. 5. Simulated time-varying signals at receiving end when Rytov variance is (a) 0.1, (b) 0.3, (c) 0.5 and (d) 0.7, respectively

Fig. 6. Threshold and judgment based on SVM detection algorithm for PAM-4 signals. (a) σR2=0.1; (b) σR2=0.3; (c) σR2=0.5; (d) σR2=0.7

Fig. 7. BER curves when Rytov variance is (a) 0.1, (b) 0.3, (c) 0.5 and (d) 0.7, respectively

Fig. 8. BER curves change with atmospheric turbulence intensity