Mobile Ad-hoc Networks (MANETs) are susceptible to various security threats due to their decentralized and dynamic nature. Among these threats, Distributed Denial of Service (DDoS) and sibling attacks pose significant challenges to the integrity and availability of network services. This paper presents a novel approach for securing MANETs against DDoS and sibling attacks through a robust Intrusion Prevention System (IPS) integrated with a secured routing mechanism. The proposed methodology leverages residual transfer learning to adapt a pre-trained model for intrusion detection to the MANET environment, enhancing its effectiveness in identifying and mitigating attacks. The problem of securing MANETs against DDoS and sibling attacks is exacerbated by the lack of centralized infrastructure and the dynamic topology of the network. Traditional security mechanisms designed for wired networks are often ineffective in MANETs due to their reliance on centralized control and communication. This research addresses this gap by proposing an IPS solution tailored specifically for MANETs, capable of detecting and preventing attacks without relying on centralized coordination. By utilizing residual transfer learning, the proposed methodology addresses the challenge of limited labeled data in the MANET domain. Transfer learning enables the adaptation of knowledge from a pre-trained model on non-MANET data to improve the performance of intrusion detection in MANET environments. The integration of the IPS with a secured routing approach ensures that detected attacks are efficiently handled within the network, minimizing their impact on performance and ensuring continued operation. Experimental results demonstrate the effectiveness of the proposed approach in mitigating DDoS and sibling attacks in MANETs. The integrated solution achieves high detection rates while minimizing false positives, thereby enhancing the security and resilience of MANETs against evolving threats.