TY - GEN
T1 - Tales from the C130 Horror Room
T2 - 1st ACM International Workshop on the Engineering of Reliable, Robust, and Secure Embedded Wireless Sensing Systems, FAILSAFE 2017
AU - Marfievici, Ramona
AU - McGibney, Alan
AU - Corbalán, Pablo
AU - Rea, Susan
AU - Rojas, David
AU - Pesch, Dirk
PY - 2017/11/5
Y1 - 2017/11/5
N2 - An important aspect of the management and control of modern data centers is cooling and energy optimization. Airflow and temperature measurements are key components for modeling and predicting environmental changes and cooling demands. For this, a wireless sensor network (WSN) can facilitate the sensor deployment and data collection in a changing environment. However, the challenging characteristics of these scenarios, e.g., temperature fluctuations, noise, and large amounts of metal surfaces and wiring, make it difficult to predict network behavior and therefore network planning and deployment. In this paper we report a 17-month long deployment of 30 wireless sensor nodes in a small data center room, where temperature, humidity and airflow were collected, along with RSSI, LQI, and battery voltage. After an initial unreliable period, a connectivity assessment performed on the network revealed a high noise floor in some of the nodes, which together with a default low CCA threshold triggered no packet transmissions, yielding a low PDR for those nodes. Increasing the CCA setting and relocating the sink allowed the network to achieve a reliability of 99.2% for the last eight months of the deployment, therefore complying with the project requirements. This highlights the necessity of using proper tools and dependable protocols, and defining design methodologies for managing and deploying WSNs in real-world environments.
AB - An important aspect of the management and control of modern data centers is cooling and energy optimization. Airflow and temperature measurements are key components for modeling and predicting environmental changes and cooling demands. For this, a wireless sensor network (WSN) can facilitate the sensor deployment and data collection in a changing environment. However, the challenging characteristics of these scenarios, e.g., temperature fluctuations, noise, and large amounts of metal surfaces and wiring, make it difficult to predict network behavior and therefore network planning and deployment. In this paper we report a 17-month long deployment of 30 wireless sensor nodes in a small data center room, where temperature, humidity and airflow were collected, along with RSSI, LQI, and battery voltage. After an initial unreliable period, a connectivity assessment performed on the network revealed a high noise floor in some of the nodes, which together with a default low CCA threshold triggered no packet transmissions, yielding a low PDR for those nodes. Increasing the CCA setting and relocating the sink allowed the network to achieve a reliability of 99.2% for the last eight months of the deployment, therefore complying with the project requirements. This highlights the necessity of using proper tools and dependable protocols, and defining design methodologies for managing and deploying WSNs in real-world environments.
KW - Low-power Wireless Communications
KW - Wireless Sensor Networks
UR - https://www.scopus.com/pages/publications/85041638445
U2 - 10.1145/3143337.3143343
DO - 10.1145/3143337.3143343
M3 - Conference proceeding
AN - SCOPUS:85041638445
T3 - FAILSAFE 2017 - Proceedings of the 1st ACM International Workshop on the Engineering of Reliable, Robust, and Secure Embedded Wireless Sensing Systems, Part of SenSys 2017
SP - 24
EP - 31
BT - FAILSAFE 2017 - Proceedings of the 1st ACM International Workshop on the Engineering of Reliable, Robust, and Secure Embedded Wireless Sensing Systems, Part of SenSys 2017
A2 - Eskicioglu, Rasit
PB - Association for Computing Machinery, Inc
Y2 - 5 November 2017
ER -