In this paper, a reconfigurable mesh-tree with a disjoint hierarchical routing protocol for the Bluetooth sensor network is proposed. First, a designated root constructs a tree-shaped subnet and propagates parameters k and c in its downstream direction to determine new roots. Each new root asks its upstream master to start a return connection to convert the first tree-shaped subnet into a mesh-shaped subnet. At the same time, each new root repeats the same procedure as the designated root to

more »

... ld its own tree-shaped subnet, until the whole scatternet is formed. As a result, the reconfigurable mesh-tree constructs a mesh-shaped topology in one densely covered area that is extended by tree-shaped topology to other sparsely covered areas. To locate the optimum k layer for various sizes of networks, a peak-search method is introduced in the designated root to determine the optimum mesh-tree configuration. In addition, the reconfigurable mesh-tree can dynamically compute the optimum layer k when the size of the network changes in the topology maintenance phase. In order to deliver packets over the mesh-tree networks, a disjoint hierarchical routing protocol is designed during the scatternet formation phase to efficiently forward packets in-between the mesh-subnet and the tree-subnet. To achieve the energy balance design, two equal disjoint paths are generated, allowing each node to alleviate network congestion, since most traffic occurs at the mesh-subnet. Simulation results show that the joint reconfigurable method and routing algorithm generate an efficient scatternet configuration by achieving better scatternet and routing performance than BlueHRT (bluetooth hybrid ring tree). Furthermore, the disjoint routing with rechargeable battery strategy effectively improves network lifetime and demonstrates better energy efficiency than conventional routing methods. Energies 2016, 9, 338 2 of 18 can be used in several applications, such as home networking, environment monitoring and sensor or ad hoc networks. With centralized infrastructures, many message delivery services are distributed via active Internet connection. Using smartphone applications, individual users can conduct packet transmissions to interested users over Bluetooth ad hoc wireless connections, while methods for sharing information over delay-tolerant networks (DTN) are being developed [4] in order to achieve social routing over Bluetooth low energy (BLE) multi-hop scatternets. Increasingly, many BLE data applications are emerging in our daily lives, such as personal area networks, body area networks, sensor networks, etc. However, these applications pose some important research challenges, such as connecting BLE devices to share information between users, and forming the desired network configurations for various types of network applications. Bluetooth wireless technology makes it possible to transmit time-constraint video/audio in mobile and pervasive environments. Most wireless sensor network (WSN) applications are event-driven. The main WSN architecture has a multi-level network topology with a large number of sensor nodes deployed within a geographic area, often communicating to an external network through a gateway node. The information communication between the sensor nodes and the gateway node is a multi-hop architecture to extend the network coverage. Typical WSN applications with such Bluetooth topology include: video streaming transmission [5], health monitoring [6] and data collection [7] . However, the performance of WSNs is constrained by limited battery capacity. To date, several approaches have been introduced to reduce the energy consumption and correspondingly maximize the lifetime of WSNs, such as data aggregation, green routing and sleep scheduling mechanisms. Among them, energy harvesting (EH) technology [8] is very promising, due to the unlimited energy supply provided by power sources such as solar power. Therefore, in order to efficiently forward packets and manage harvesting energy for a rechargeable Bluetooth sensor network, this study proposes a reconfigurable mesh-tree with a disjoint hierarchical routing protocol. In contrast to our prior work [9] , the size of the mesh-subnet is fixed and predefined by the constant k value regardless of the network size. However, the reconfigurable mesh-tree determines the variant k instead of the constant k layer in the mesh-subnet for the various sizes of networks. In addition, the reconfigurable mesh-tree can dynamically compute the optimum layer k when the size of the network changes in the topology maintenance phase. As a result, the reconfigurable mesh-tree constructs a desired mesh-shaped topology in one densely covered area that is extended by several tree-shaped topologies to other sparsely covered areas. Thus far, most scatternet formation methods partition networks by collecting complete topology information [10-12], thereby introducing considerable computation or communication overheads. To mitigate topology formation overhead, a peak-search method is proposed with partial topology information. The peak-search method designs three blocks: the connection link, the hop length and the optimum decision blocks. The connection link block calculates the total number of links, the hop length block computes the average query hop and the optimum decision block uses a decision-making criterion to locate the desired k layer for the mesh-subnet. With the packet forwarding in the mesh-tree topology, most of the sensor traffic loads pass through the mesh-subnet to the sink node. From the energy consumption point of view, the nodes nearer the sink forward most of the packets, and deplete their batteries more rapidly than other nodes, and thus become an energy hole in a network. To balance the traffic load and mitigate the energy hole problem in the mesh-subnet, two equal disjoint paths are designed for each node forwarding packets to the sink. The design goal of the routing algorithm is to balance the energy consumption for hot-spot nodes, and to deliver packets from one subnet to all other subnets effectively. Thus, by simultaneously harvesting renewable energy, the network lifetime of the hot-spot nodes can be prolonged by the two equal paths routing design. The remainder of this paper is arranged as follows: Section 2 reviews current Bluetooth scatternet formation and routing protocols. In Section 3, the reconfigurable mesh-tree concept and the formation Energies 2016, 9, 338 3 of 18 algorithm are presented. In Section 4, the peak-search method to achieve the desired configuration is described in detail. In Section 5, the hierarchical routing protocol for the reconfigurable mesh-tree topology is presented. In Section 6, computer simulations are used to demonstrate the scatternet, routing and energy performances. Finally, conclusions are offered in Section 7.