Reliable data transfer seems a quite challenging task in Underwater Wireless Sensor Networks (UWSN) in comparison with Terrestrial Wireless Sensor Networks due to the peculiar attributes of UWSN communication. However, the reliable data transmission in UWSN is very limited. Yet, there is a way to achieve reliable data transfer metrics through the design of routing protocols by considering the exceptional features of UWSN communications. With this aim, we propose two schemes with multiple sinks-based network architecture: Anchor Nodes assisted Cluster-based Routing Protocol (ANCRP) to achieve reliable data transfer metrics and Void Handling technique in ANCRP (VH-ANCRP) to cope with the local maximum nodes. For which, the network space is divided into small cubes to form clusters. Then, each cube is assigned with an anchor node as a cluster head (CH). All cluster heads are supposed to be anchored at the centroid of a cube via a string, while source nodes are randomly distributed. In ANCRP, the source nodes are liable to send the sensed data to their designated CH. The CH transmits the sensed data to the next-hop CH and continues this procedure till the successful delivery of the data packets at the surface sinks. In VH-ANCRP, a void handling technique of making the ad-hoc CH is used by the void nodes to reconnect with the network operations. We perform extensive simulations in NS3 to validate our schemes. The simulation outcomes expel that both proposed schemes have improved the network performance when compared with the baseline schemes. INDEX TERMS Underwater wireless sensor networks, reliable data transfer, routing protocol, anchor nodes, clustering technique, void handling. I. INTRODUCTION Currently, Underwater Wireless Sensor Networks (UWSNs) are gaining pivotal considerations in both industrial and academia sector because of their wide and comprehensive implementation areas, such as resource exploration, navigation assisting, military surveillance, calamity preventions, etc. [1]. The underwater WSNs also assist in finding the unexplored underwater resources and aquatic data collection with the help of different computational intelligence approaches [2]. The underwater wireless sensor nodes are supposed to be deployed sparsely from surface-layer to seabedlayer for fetching the data from the underwater harsh environments by using an acoustic modem [3], [4]. Underwater communication through optical signals is not feasible due to absorption loss and rapid attenuation. Thus, underwater communication is carried out in acoustic signals [5]. The