Peer-to-Peer Netw. Appl. DOI 10.1007/s12083-015-0360-0 CLEVER: Cluster-based Energy-aware Virtual Ring Routing in randomly deployed wireless sensor networks Ghofrane Fersi 1 · Wassef Louati 1 · Maher Ben Jemaa 1 Received: 30 September 2014 / Accepted: 20 April 2015 © Springer Science+Business Media New York 2015 Abstract Energy-aware routing is an important remedy to face the quick failure of energy-constrained nodes in Wire- less Sensor Networks. Network clustering with electing energy-powerful nodes as cluster heads is a perfect solution. However, such clustering requires ideal nodes placement to afford best performances. Manual nodes placement is not always possible, the sensors can be randomly deployed. In such networks, the cluster heads cannot always communi- cate directly. In this paper, we present a novel clustering strategy for randomly deployed heterogeneous sensors, in which a cluster is defined as a set of energy-powerful nodes placed at the range of each other. The proposed protocol, called CLEVER (Cluster-based Energy-aware Vir- tual Ring Routing), uses virtual identity-based routing for intra and inter-cluster communications. The experimental results show that CLEVER increases drastically the network lifetime and optimizes efficiently sensors energy. Keywords Wireless sensor networks · Distributed Hash Tables · Energy-aware routing · Clusters  Ghofrane Fersi ghofrane.fersi@redcad.org Wassef Louati wassef.louati@redcad.org Maher Ben Jemaa maher.benjemaa@enis.rnu.tn 1 Research Laboratory of Development and Control of Distributed Applications (ReDCAD), Department of Computer Science and Applied Mathematics, National School of Engineers of Sfax, University of Sfax, BP 1173-3038 Sfax, Tunisia 1 Introduction Wireless Sensor Networks (WSNs) are made up of a large number of sensors that have sensing, processing and communication functionalities [1, 2]. The main role of these sensors is to collaborate with each other in order to supervise given phenomena and to send results to base stations. WSNs are facing a lot of challenges such as energy optimization, fault tolerance, network coverage, security. Energy optimization remains one of the most important challenges since it is directly related to the network life- time. A lot of researches [3] have been proposed in order to face this challenge. Most of these researches, assume that the network is homogeneous, that is all sensor nodes have the same sensing and processing capabilities and the same initial amount of energy. However, in reality, the WSN is made up of heterogeneous sensors having different features and various modalities. Taking into account such diversity can improve the network performance and specifically the network lifetime. One of the most trivial solutions in the case of hetero- geneous WSN, is to partition the network into clusters and to elect energy powerful nodes as cluster heads. In each cluster, energy weak nodes send their messages to the clus- ter head which in turn send the fused messages to the base station directly or by using other cluster heads as interme- diate nodes [4, 5]. Such solutions suite deterministic WSN deployment cases, where cluster heads can be placed at the range of each other. However, deterministic deployment is not always feasible. For example, in the case of natural dis- asters, or hostile supervision, the supervised area cannot be accessed directly. Sensors are scattered randomly and they should form autonomously their own network without any humain intervention. The cluster heads can be placed so