Ransomware is a form of malicious software which when deployed encrypts or locks the files and demands a ransom to have the files decrypted and accessible. In today’s digital world, devices connected to the network are vulnerable to ransomwares. Signature based methods are used to detect known ransomwares at an early stage to minimize the damage. This type of detection involves a predefined repository of static signatures which block the ransomwares based on the signatures in the repository. Fuzzy hashing method and Clam Av method for signature based ransomware detection have been proposed in literature. However, these methods raised false negatives and false positives respectively. Also, the speed of comparison in fuzzy hashing method was a limitation. In this paper an Ant Colony Optimization (ACO) based approach for filtering ransomwares by matching the incoming signature hash value with the signature database is proposed. Termed Ant Colony Optimization based Light weight Binary Search for Signature based Ransomware detection (ACOLBSR) algorithm, the ant agent finds the search space in the signature database. If a search space does not exist in the signature database, the incoming file is logged. If the search space exists in the database, the ant agent employs a binary search to match the incoming signature hash value with the signatures in the search space. If a match is found, the file is accepted or blocked if it is a goodware or ransomware respectively. If the incoming signature does not match with the signatures in the search space, the file is logged. The advantage of ACOLBSR algorithm is that the ransomwares are strictly filtered based on the signatures in the database. The computational complexity of ACOLBSR algorithm is less compared to binary search and sequential search. Also, it is shown that ACOLBSR algorithm do not generate false positives when compared to ClamAV, SplitScreen, fuzzy hashing and classification based approaches. Experimental results comparing the performance of the ACOLBSR scheme with the binary search scheme, sequential search scheme and classification based approaches are presented.