In this paper, given a decentralized multi-agent system, a protocol for reaching trust based consensus is examined as it plays a great role in control. Depending on different factors (e.g. agent's specifications, environment, communication, etc.), the agent with the greatest trust value should take the leading role of the group at that moment. Thus, a system hierarchy is established based on trust values of each agent and they are being updated in every step of the algorithm. The values are obtained by exploiting the current agent's accomplishment (local observation) and observations received from the other agents in its neighbourhood. Considering the fact that not all agents have to agree separately on the trust value, an adaptation control is employed to negotiate and find the best solution together in the end. In general, invariable topology of data exchange is used more often because of its simplicity. In this paper, a trust based consensus is trying to be reached having a system with variable topology since it 'imitates' a real system more credibly. Furthermore, trustworthiness of the binary protocol is examined by simulating the algorithm in MATLAB, just like the influence of a variable measurement error (a difference between measured data and data obtained in communication with other agents) on protocol execution. Moreover, the algorithm is adapted and tested in an underwater simulator that represents a real system. The findings of this study strongly suggest the binary protocol in multi-agent systems with variable topology because it enables reaching the consensus in trust values and makes a self-organizing cooperative control in the system possible.