{"id":7740,"date":"2026-09-25T23:34:53","date_gmt":"2026-09-25T15:34:53","guid":{"rendered":"https:\/\/people.utm.my\/murtadha\/?page_id=7740"},"modified":"2026-09-25T23:34:53","modified_gmt":"2026-09-25T15:34:53","slug":"manet-trust-prediction","status":"publish","type":"page","link":"https:\/\/people.utm.my\/murtadha\/research-areasinterest\/manet-trust-prediction\/","title":{"rendered":"MANET Security and Trust Prediction: Study Description"},"content":{"rendered":"
\n

Study description<\/p>\n

MANET Security and Trust Prediction: Study Description<\/h1>\n
\n

About this page<\/h2>\n

This page describes what three published IEEE Access articles and one correction examine, for readers who want a plain-language orientation to the studies. It does not report, compare or rank the performance of any method. Readers who need results should consult the articles listed under Publications.<\/p>\n<\/div>\n

Short answer<\/h2>\n

The studies ask whether a trust value predicted for each node can help a mobile ad hoc network (MANET) choose routes when some nodes drop the packets they are asked to forward. One research article predicts node trust with a Long Short-Term Memory (LSTM) model and uses the predictions within location-aided routing. A second research article examines federated learning (FL) and multiobjective optimisation in a trust-aware routing framework. A third article is a survey of intrusion-detection approaches for black-hole and gray-hole attacks in MANETs. The two research articles evaluate their methods in simulation. This page describes the studies and does not report their measured results.<\/p>\n

\n
\n

The research problem<\/h2>\n

A mobile ad hoc network (MANET) is a group of wireless mobile nodes that communicate with one another without relying on fixed infrastructure or central management. Because nodes forward each other’s packets, a malicious node can disrupt network operation, for example by dropping the packets it receives. The articles start from the observation that conventional routing protocols often lack mechanisms to deal with such nodes. The research problem is therefore how a node’s trustworthiness might be estimated from observed behaviour and used when routes are selected.<\/p>\n<\/div>\n

\n

What node trust means<\/h2>\n

In the LSTM article, node trust is an article-defined score intended to reflect how reliably a node forwards packets. The simulation begins from a neutral trust value and groups lower and higher values into trust categories. The article estimates trust from features of observed node behaviour, such as packet forwarding and loss ratios, link quality and neighbour stability. This is the article\u2019s working definition for its simulations.<\/p>\n<\/div>\n

\n

Black-hole and gray-hole attacks<\/h2>\n

A black-hole attack occurs when a malicious node claims to offer the shortest route to a destination, attracts traffic, and then discards every packet it receives. In a gray-hole attack, a malicious node drops packets selectively, which makes it harder to detect. In the LSTM article’s simulations, malicious nodes are set to drop data packets.<\/p>\n<\/div>\n<\/div>\n

What the studies examine<\/h2>\n
\n

LSTM trust prediction and location-aided routing<\/h3>\n

The LSTM article proposes a routing protocol called LSTMT-LAR, in which each node predicts the trust of other nodes and the predictions inform route selection. An LSTM is a type of recurrent neural network suited to data that arrive in sequence, such as behaviour observed over time. Location-aided routing (LAR) is a protocol that uses node location information to limit the search for routes. The prediction uses a model of 13 behavioural features collected for each pair of nodes. Trust values used for training are produced inside the simulation: a behaviour model generates node activity, features are extracted, and a labelling function assigns trust scores.<\/p>\n<\/div>\n

\n

Federated learning and multiobjective optimisation<\/h3>\n

The federated learning article presents a framework called FLSTMT-LAR. In federated learning (FL), models are updated collaboratively; in this framework, nodes exchange model updates with neighbouring nodes to update their local trust models. Multiobjective optimisation is used to balance trust against other routing objectives, and the article compares two algorithms, NSGA-II and NSGA-III. LSTMT-LAR is one of the protocols compared in this article.<\/p>\n<\/div>\n

\n

Survey of MANET intrusion-detection approaches<\/h3>\n

The survey article reviews published intrusion-detection approaches for black-hole and gray-hole attacks in MANETs. It covers MANET characteristics, the two attack types, machine-learning methods including LSTM, optimisation approaches such as metaheuristics, game theory and reinforcement learning, variants of location-aided routing, and federated learning. It organises and discusses existing work rather than reporting a new experiment. It also discusses limitations of LSTM-based approaches and open research directions.<\/p>\n<\/div>\n

How the studies were evaluated<\/h2>\n

Both research articles evaluate their methods by simulation. The LSTM article states that its simulation environment was implemented in Python. The FL article describes its evaluation as using simulated environments. The LSTM article evaluates four scenarios that vary node mobility (static or random walk), the number of nodes and the proportion of malicious nodes. The FL article evaluates five scenarios covering static networks, random-walk mobility, increasing numbers of nodes and a comparison of optimisation algorithms. Both articles compare their protocols with other location-aided routing variants and with flooding and random routing. This page gives design descriptions only and no performance figures; the articles report numerical results for several network measures, and those results are not reproduced or compared here.<\/p>\n

\n

What the publications do not establish<\/h2>\n

This page does not assess whether any described method is effective, and it does not compare or rank methods. Results from simulation do not by themselves describe how a method would behave in a physical network. This page does not assess how well the trust model predicts trust values, and details of the training setup are given in the articles and not summarised here. This page does not assess privacy properties of the federated learning design.<\/p>\n<\/div>\n

Publications<\/h2>\n
\n

1. Research article (simulation-based evaluation). LSTM trust prediction and location-aided routing.<\/strong><\/p>\n

Saad Mohsen Hassan, Mohd Murtadha Mohamad and Farkhana Binti Muchtar (2025). Enhancing MANET Security Through Long Short-Term Memory-Based Trust Prediction in Location-Aided Routing Protocols. IEEE Access, 13, 120142\u2013120168.<\/p>\n

DOI: https:\/\/doi.org\/10.1109\/ACCESS.2025.3572619<\/a><\/p>\n

Licence stated on the article: CC BY 4.0.<\/p>\n<\/div>\n

\n

2. Research article (simulation-based evaluation). Federated learning and multiobjective optimisation.<\/strong><\/p>\n

Saad M. Hassan, Mohd Murtadha Mohamad, Farkhana Binti Muchtar and Firoz Bin Yusuf Patel Dawoodi (2024). Enhancing MANET Security Through Federated Learning and Multiobjective Optimization: A Trust-Aware Routing Framework. IEEE Access, 12, 181149\u2013181178.<\/p>\n

DOI: https:\/\/doi.org\/10.1109\/ACCESS.2024.3505236<\/a><\/p>\n

Licence stated on the article: CC BY 4.0.<\/p>\n<\/div>\n

\n

3. Survey article.<\/strong><\/p>\n

Saad M. Hassan, Mohd Murtadha Mohamad and Farkhana Binti Muchtar (2024). Advanced Intrusion Detection in MANETs: A Survey of Machine Learning and Optimization Techniques for Mitigating Black\/Gray Hole Attacks. IEEE Access, 12, 150046\u2013150090.<\/p>\n

DOI: https:\/\/doi.org\/10.1109\/ACCESS.2024.3457682<\/a><\/p>\n

Licence stated on the article: CC BY-NC-ND 4.0.<\/p>\n

\n

Erratum, presented as a note to the survey article (3).<\/em><\/p>\n

Saad M. Hassan, Mohd Murtadha Mohamad and Farkhana Binti Muchtar (2025). Corrections to \u201cAdvanced Intrusion Detection in MANETs: A Survey of Machine Learning and Optimization Techniques for Mitigating Black\/Gray Hole Attacks\u201d. IEEE Access, 13, 137850.<\/p>\n

DOI: https:\/\/doi.org\/10.1109\/ACCESS.2025.3595635<\/a><\/p>\n

The correction states that reference [4] of the survey was retracted and is replaced in the reference list by another reference. It describes this change only.<\/p>\n<\/div>\n<\/div>\n

\n

Code and data availability<\/h2>\n

No public code or data repository was identified in the sources reviewed. \u201cSources reviewed\u201d here means the three published articles, the erratum and the public web pages examined for this page. This sentence does not state that code or data do not exist.<\/p>\n<\/div>\n

Related research<\/h2>\n