Electroencephalogram (EEG) contains important physiological information that can reflect the activity of human brain so that it is useful for epileptic seizure detection and epilepsy diagnosis. In this paper, we develop a novel unified framework for real-time monitoring of EEG for epileptic seizure prediction with minimum number of electrodes for wearable application without any prior knowledge. This research work uses Principal Component Analysis (PCA) for ranking the highest contribution of channels during seizure period to reduce the number of EEG scalp electrodes. CHB-MIT data has average 23 channels for each patient, but obtained results of our study show that average five to six channels are enough to get good sensitivity with less False Prediction Rate (FPR) per hour. In this research work, different channels combinations in the term of accuracy, sensitivity and FPR/hr have been analyzed and the result obtained 92.73% sensitivity and 0.073 FPR/hr for only 10 channels.