Caching aims to store data locally in some nodes within the network to be able to retrieve the contents in shorter time periods. However, caching in the network did not always consider secure storage (due to the compromise between time performance and security). In this paper, a novel decentralized secure coded caching approach is proposed. In this solution, nodes only transmit coded files to avoid eavesdropper wiretapping and protect the user contents. In this technique random vectors are used to combine the contents using XOR operation. We modeled the proposed coded caching scheme by a Shannon cipher system to show that coded caching achieves asymptotic perfect secrecy. The proposed coded caching scheme significantly simplifies the routing protocol in cached networks while it reduces over-caching and achieves a higher throughput capacity compared to uncoded caching in reactive routing. It is shown that with the proposed coded caching scheme any content can be retrieved by selecting a random path while achieving asymptotic optimum solution. We have also studied the cache hit probability and shown that the coded cache hit probability is significantly higher than uncoded caching. A secure caching update algorithm is also presented.