Wireless sensor network for temperature monitoring and cooling optimization in data centers

The trend for Cloud computing is leading to the increasing of data centers due to the necessity to store a big amount of information, which is rising over time. Companies such as Amazon, Google, Microsoft, Facebook and Apple have decided to offer information stored in the Internet Cloud, ensuring faster and more efficient services to the users. Nowadays, data centers consume about 2% of electricity worldwide and contribute 0.3% to global carbon emissions. The cooling infrastructure is needed to ensure that IT resources operate within a safe range of temperatures, ensuring reliability. However, cooling infrastructure consumes up to 40% of the power consumption of the data center. Thus, efficient cooling systems are necessary to reduce the server failures due to the increasing temperature for IT equipment and energy saving. The choice to exploit wireless sensors is due to the fact they have different advantages, such as not requiring to make changes to the existing infrastructure, being easy to deploy and relocate, and being relatively low cost. As we are working with critical infrastructures, and cooling control is one of the most critical operation decisions, the primary objective it has been to guarantee safety in operation. For that reason, formal verification methods and real-time analysis techniques have been applied to the design. The system requirements and the system design as finite state machines have been defined and verified with the model checker tool called New Symbolic Model Verifier (NuSMV). Two FSMs have been designed: one to manage the reading and sending data from the sensors, and another one to manage the sleeping mode when the sensors are not required to operate. Additionally, the implementation and the power consumption were also analyzed.