We propose the RLizard key encapsulation mechanism (KEM), whose security depends on the ring learning with errors and ring learning with rounding problems. Because RLizard operates on a special type of ring, it is more efficient in terms of both the clock cycles required for key generation and the key size compared with the original Lizard scheme. To demonstrate the superiority of the proposed method over other well-known KEMs, we compared their performances in the 32-bit ARM Internet of Things (IoT) environment. The performance analysis showed that the RLizard KEM requires the fewest clock cycles for key generation, encapsulation, and decapsulation when the parameters are set to support a security level comparable with that of AES-128. In summary, the RLizard KEM is expected to be used for secure communication and authentication between IoT endpoint devices, whose computational power is generally limited.