Wireless sensor networks which form part of the core for the Internet of Things consist of resource constrained sensors that are usually powered by batteries. Therefore, careful energy awareness is essential when working with these devices. On the other hand, the presence as well as the absence of security features implemented in resource constrained sensors can have negative effects on their energy consumption. Indeed, the introduction of security techniques such as authentication and encryption, in order to ensure confidentiality and integrity of data, can place higher energy load on the sensors. However, the absence of security protection could give room for energy-drain attacks such as denial-of-sleep attacks which has a higher negative impact on the life span (availability) of the sensors than the presence of security techniques. This paper focuses on denial-of-sleep attacks by simulating three Media Access Control (MAC) protocols – Sensor-MAC, Timeout-MAC and TunableMAC – under different network sizes. We evaluate, compare, and analyse the received signal strength and the link quality indicators for each of these protocols. The results of our simulation provide insight into how these parameters can be used to detect a denial-of-sleep attack. Finally, we propose a novel architecture for tackling denial-of-sleep attacks by propagating relevant knowledge via intelligent agents.