Wireless sensor networks (WSNs) have many limitations in terms of their computational capability, memory, and battery power. In addition, the sensor nodes use insecure wireless communication channels. Thus, WSNs are vulnerable to many types of attacks such as jamming, sinkhole, wormhole, Sybil, and selective forwarding. Among all these attacks, detecting a wormhole attack is very difficult in mobility-based WSNs. In these WSNs, two genuine nodes that are located far apart can become one-hop neighbours for valid reasons. However, the same scenario might indicate the presence of a wormhole attack instead. Differentiating genuine nodes from malicious nodes is a highly challenging task. Our detection method is based on the concept of the rate of change of neighbouring nodes and the length of an alternative path between two nodes. The proposed method does not require any additional hardware such as synchronised clocks or timing information, GPS, or cryptographic methods that require large amounts of computational power. The simulation results indicate that our method has good detection accuracy.