Electronic Theses and Dissertation

Mobile Ad Hoc Network (MANET) merupakan jaringan komunikasi nirkabel yang dapat beroperasi tanpa infrastruktur tetap dan banyak dimanfaatkan dalam situasi darurat seperti komunikasi militer dan penanggulangan bencana. Namun, sifatnya yang dinamis dan terdesentralisasi membuat MANET rentan terhadap serangan keamanan, salah satunya adalah serangan blackhole yang menyebabkan hilangnya data dan penurunan kualitas layanan jaringan. Penelitian ini penerapan model Long Short-Term Memory (LSTM) untuk mendeteksi serangan blackhole dengan memanfaatkan parameter jaringan seperti throughput, delay, jitter, dan packet loss. Dataset untuk pelatihan dan pengujian model dikembangkan melalui simulasi jaringan MANET menggunakan NS-3. Hasil penelitian menunjukkan bahwa model LSTM mampu mencapai akurasi pelatihan dan validasi di atas 99,5%. Pada pengujian data baru, model berhasil mendeteksi data serangan blackhole sebesar 80,49% dan data normal sebesar 19,51%. Temuan ini membuktikan bahwa LSTM mampu mengenali pola anomali yang ditimbulkan oleh serangan blackhole secara efektif. Selain itu, penelitian ini juga menganalisis dampak serangan blackhole yang menyebabkan throughput menurun tajam, peningkatan packet loss secara signifikan, serta mempengaruhi delay dan jitter jaringan.

Kata Kunci: MANET, AODV, Blackhole Attack, LSTM, NS-3.

Mobile Ad Hoc Network (MANET) is a wireless communication network that can operate without fixed infrastructure and is widely utilized in emergencies such as military communications and disaster response. However, its dynamic and decentralized nature makes MANET vulnerable to security threats, one of which is the blackhole attack, where malicious nodes cause data loss and degrade the quality of network services. This study implements the Long Short-Term Memory (LSTM) model to detect blackhole attacks by utilizing network parameters such as throughput, delay, jitter, and packet loss. The dataset for training and testing the model was generated through MANET network simulations using NS-3. The research results show that the LSTM model achieved training and validation accuracy exceeding 99.5%. In testing on new data, the model successfully detected blackhole attack data at 80.49% and normal data at 19.51%. These findings demonstrate that LSTM can effectively recognize anomaly patterns caused by blackhole attacks. Furthermore, this study also analyzes the impact of blackhole attacks, which significantly decrease throughput, increase packet loss, and affect network delay and jitter. Keywords: MANET, AODV, Blackhole Attack, LSTM, NS-3.