Attribute-based keyword search (ABKS) achieves privacy-preserving keyword search and fine-grained access control over encrypted data in the cloud. However, existing ABKS schemes cannot be directly applied for resource-constrained (such as Internet of Things) devices due to heavy computation overhead. In addition, identifying the malicious user who misuses the secret key is difficult if more than one user is having the same set of attributes. Furthermore, user revocation and attribute revocation are two important challenges in real-world applications. To address these challenges, this article proposes a FELT-ABKS: fog-enabled lightweight traceable ABKS over encrypted data by using ciphertext-policy ABKS to realize keyword search and fine-grained access control. FELT-ABKS achieves minimal computation cost at end users by transferring maximum computation to fog nodes. Furthermore, FELT-ABKS traces the malicious users who misuse their secret key. Besides, it supports user revocation and attribute revocation. The security analysis proves that FELT-ABKS is secure against the chosen keyword attack, chosen-plaintext attack, and modify secret key attack. Finally, experiments demonstrate that FELT-ABKS is lightweight and feasible.