Communication systems in aviation tend to focus on safety rather than security. Protocols such as Automatic Dependent Surveillance-Broadcast (ADS-B) use plain-text, unauthenticated messages and, therefore, open to various attacks. The open and shared nature of the ADS-B protocol makes its messages extremely vulnerable to various security threats, such as jamming, flooding, false information, and false Squawk attacks. To handle this security issue in the ADS-B system, a state-of-theart dataset is required to train the ADS-B system against these attacks using machine learning algorithms. Therefore, we generated the dataset with four new attacks: name jumping attack, false information attack, false heading attack, and false squawk attack. After the dataset generation, we performed some data pre-processing steps, including removing missing values, removing outliers from data, and data transformation. After pre-processing, we applied three machine learning algorithms. Logistic regression, Naive Bayes, and K-Nearest Neighbor (KNN) are used in this paper. We used accuracy, precision, recall, F1-Score, and false alarm rate (FAR) to evaluate the performance of machine learning algorithms. KNN outperformed Naive Bayes and logistic regression algorithms in terms of the results. We achieved 0% FAR for anomaly messages, and for normal ADS-B messages, we achieved 0.10% FAR, respectively. On average more than 99.90% accuracy, precision, recall, and F1-score are achieved using KNN for both normal and anomaly ADS-B messages.