Wireless mesh networks are self-forming, dispersed networks of communication nodes. In a wireless mesh network, the nodes act as the network's infrastructure rather than connecting to a central access point or router. The range and coverage of a network may be increased by the transmission of data packets from one node to another [15, 16]. The problem of gateway selection in wireless mesh networks (WMNs) has been studied extensively in the past [8–15]. Wireless transceivers are often installed in each node in a wireless mesh network so that they may exchange data with one another. Together, these nodes dynamically construct and maintain connections throughout the network, making it resilient to the failure or removal of individual nodes [17, 18].
The Multi-Rate Multi-Carrier (MRMC) design is widely recognized as a viable method for improving system performance, however the authors have not taken it into consideration. As a result, the studies under consideration have not been compared to the approach described in this study. Distance to gateway nodes has been the primary focus of prior studies [16], as this has been shown to be a major cause of both lower performance and an unequal distribution of traffic load among gateway nodes. This discrepancy arises when certain gateway nodes are underutilized while others are clogged with excessive traffic. If there is a shorter connection between the source node and the gateway node, but it has a significant probability of interference and packet loss, then the hop count metric may not be the best option. In [11], the authors described a proactive algorithm for centralized gateway selection in multi-hop wireless networks using a single radio and a single channel (SRSC). In this investigation, the hop count of the wireless hosts was similar to that of the gateways. Each host used a cost function depending on load balancing to identify its default gateway. The interference from wireless channels close to the gateways was considered with traffic density and load balancing in this analysis. This research shows that when two cost functions are considered together, as in the suggested system, gateway throughput improves. The gateways in this technique were selected solely based on the hop count metric, with the use of a hop count threshold and the establishment of gateway selection criteria limiting the candidates to a smaller range of nodes. Potential drawbacks of this approach include diminished performance and an inequitable allocation of traffic. The high network cost, poor scalability, and unreliability of centralized algorithms plainly make them inferior to their distributed counterparts. Several tests, like the one reported in [17], have seen heuristic and centralized gateway switching without optimization for better gateways and paths. What's more, the aforementioned studies didn't use learning methods that take into consideration the present state of the network environment to facilitate appropriate gateway and route selection. For MRMC-enabled multi-tier wireless mobile ad hoc networks (MANET), the authors of reference [18] created a link-state and distributed approach. Their method is proactive in that it selects gateways inside each sub-network on the fly. It's impossible to have both low hop counts and low levels of interference. This research tries to strike a balance between the opposing viewpoints. Gateway nodes are separated into two groups, active gateways and backup gateways, to improve network responsibility in emergency situations. When a gateway node is added, removed, or replaced in a network, communication is temporarily disrupted until the nodes can reorganize and the pathways inside the network can reconverge. Over time, the absence of additional backup gateway nodes has also hampered the achievement of even a suitable sub-optimal solution. In order to determine which gateway nodes should be assigned to which sub-networks, the algorithm takes into account both the existing status of the network and the anticipated plan for the future relocation of nodes. The improved inner gateway routing protocol (EIGRP) served as inspiration for the ranking scheme used here, which takes into account both connection costs and congestion costs. This list, however, is used to calculate gateway costs in preparation for future sub-network allocation. Despite the importance of the path state between wireless routers and gateway nodes, some studies [17, 19, 20] only considered traffic load. In a self-organizing, self-configuring wireless mesh network (SRSC WMN), gateway load has been proposed as a criterion for selecting the best gateway. The end goal is to distribute traffic evenly throughout all of the gateways. It is not prudent to pick routes based merely on the amount of traffic at the entrance. In other cases, the gateway's load may be modest, but the pathways going there may be very susceptible to interference and packet loss. In the context of SRSC WMNs, a distributed method has been the subject of many other research [8]. This algorithm takes a preventative approach to selecting gateways, routes, and routes. In this analysis, an algorithm known as Best Gateway Selection (BGS) was used. BGS took into account a number of different variables, including gateway load and ETX, to make its final decision.
Wireless mesh networks provide the advantages of scalability, flexibility, and dependability. Since each node may communicate directly with many of its neighbors, the network can be easily expanded by adding additional nodes. Because data may be diverted to operational nodes in a mesh network, it is more resilient to disturbances and outages [19, 20].
Wireless mesh networks have the potential to improve communications in many various contexts, including homes and companies, "smart cities," industries, and even emergency situations. They allow for safe and effective communication, connection and sharing of equipment across vast distances and in challenging or variable environments [21].
Some key features that set wireless mesh networks apart are as follows:
In the event of a node failure or a change in the network's topology, a wireless mesh network may be able to self-heal and return to normal operation without human intervention. When a node goes down, the network may stay up and running with minimal disruptions provided neighboring nodes can dynamically reroute traffic through other routes [22].
Nodes in a wireless mesh network collaborate to set up and maintain the network's infrastructure decentralized from a central authority. Self-organization techniques enable nodes to autonomously configure communication with their neighbors [23, 24], as opposed to relying on human configuration.
In order to get data where it needs to go, nodes in a wireless mesh network "hop" from one another. Multi-hop communication [25, 26] allows for greater network coverage and communication between distant nodes.
Since wireless mesh networks can be set up quickly, even in locations with little to no existing infrastructure, they can provide ad hoc connectivity. The ability of individual nodes to quickly and easily set up a network via ad hoc connections makes this architecture suitable for situations when setting up a permanent network infrastructure would be problematic, such as during disaster recovery or temporary deployments [27, 28].
A wireless mesh network's coverage area and throughput may be easily increased by adding additional nodes, which can be done with relative ease. The mesh network's decentralized architecture and efficient resource utilization will allow it to handle more traffic and more devices as the network grows [29, 30].
Wireless mesh networks have a wide variety of applications, including but not limited to community networks, smart homes, outdoor wireless networks, wireless sensor networks, and Internet of Things (IoT) deployments. They provide a versatile and adaptable alternative to traditional infrastructure-based networks in scenarios when doing so would be impractical or prohibitively expensive, such as with wireless communication [31, 32].
In addition, our research systematically evaluated interference and connection quality to identify the best route to gateway nodes. However, the measurement relied only on the estimated transmission count (ETX) to predict the expected link quality (ELQ), ignoring other parameters that might have an impact on the ELQ. Expected Transmission Time (ETT) and ETX [21, 22] are routing measures that exclusively consider delivery ratio and ignore any interference that may occur between connections. This load imbalance originated from these measures' inaccurate depiction of traffic volume. The authors of reference [17] described a heterogeneous Mobile Ad hoc Network (MANET) with two layers of connectivity. To solve the problem of gateway selection using the existing Metric Link Indicator (MLI) metric, the network was outfitted with a distributed algorithm. In this analysis, the gateway's current load was evaluated using historical data and the amount of traffic that must be processed. Furthermore, this method included a technique to stop hosts from rapidly moving between gateway nodes. If a host has been using gateway i for some time and then gets an advertising from gateway j suggesting a lower load, the host may choose to change its network configuration to use gateway j instead. The load index of j also has to be less than the load index of i by at least a threshold value, L. In addition, in the aforementioned cases, a node strategically switches gateways with a fixed probability to reduce the frequency of switches [21].
It has been shown that deploying networks is NP-hard [9], [7], [22], and gateway placement in wireless networks is only one example. In paper [8] conducted by Chandra set of placement methods provided to address network capacity limits, wireless interference, fault tolerance, and fluctuating traffic needs. The asymptotic capacity of wireless multi-hop networks was investigated by Gupta and Kumar [11]. As described in references [21] and [12], several models were employed to evaluate the capacity of wireless networks. In [20] analyzed the capacity of random multi-hop multi-radio multi-channel wireless networks. In [17] used the protocol interference model to investigate the issues of flow routing and transmission scheduling within a single-channel wireless network. By exploring the potential of multi-radio multi-channel mesh networks, this study builds upon earlier attempts. Alicherry et al. [1] offer a technique based on linear programming to maximize throughput in MIMO WMNs. In this method, multipath routing, link scheduling, and static channel assignment are all handled concurrently.
All of the aforementioned studies either concentrated only on the capabilities of multi-hop wireless mesh networks (WMNs) without gateways or else assumed a static environment in which the positions of the mesh routers and gateways were fixed. Recently, Li et al. [22] conducted study in which they analyzed the unsure location of gateways inside the mesh backbone. Finding the optimal location for gateways was a priority for the team so that they could maximize throughput. As a workaround, the researchers proposed a novel grid-based gateway deployment technique based on cross-layer throughput optimization.