We develop an experimental environment to verify decentralized real-time optimization methods using publish/subscribe (pub/sub) communication with low QoS settings. Research on the distributed constraint optimization problem (DCOP) explores fundamental combinatorial optimization problems in multiagent systems. However, the implementation of solution methods using practical communication systems requires further investigation. Recent robot control platforms and frameworks for IoT devices employ pub/sub communication models for relatively simple communication. A previous study addressed this situation and investigated a variant of an anytime local solution method for DCOPs allowing for temporarily missing agents in the solution process. We experimentally investigate this approach by implementing it in an environment of small board computers with an ROS2 framework. We also describe/verify the methodology of our practical experiment.