Genetic algorithm-based optimisation of load-balanced routing for AMI with wireless mesh networks
Introduction
Smart grid [1] is an enhanced form of the existing power grid [2] that ensures availability, reliability, and efficiency. AMI [3], [4] is a gateway where the grid integrates with the consumer. The smart grid communication infrastructure comprises three major communication infrastructures [5], including (i) home area networks (HANs): short-range communication that connects a number of devices and sensors, (ii) neighbourhood area networks (NANs): medium-range communication connecting HAN with WAN, and (iii) wide area networks (WANs) that connect many NANs to a central control system. The AMI is the key NAN component collecting smart meter data from the NAN and aggregates it before sending it to the central control centre. To implement the AMI, some basic requirements are essential [6], [7]. Coordinating all the functional requirements demands a suitable communication infrastructure [8]. The AMI requires an efficient two-way communication network to support smart meters that send periodic messages carrying meter reading information. The communication network must contain an efficient topology and a routing protocol. Yasin et al. [5] summarised various technologies for smart meter communication in smart grid communication infrastructure.
Several network topologies and protocols exist to construct an AMI communication network. Packet routing is an important aspect of AMI network operations. The routing method depends on the type of communication technology used to implement it. Herein, the AMI communication network is deployed using a mesh topology. Diegio et al. [9] examined all possible routing protocols belonging to assorted network technologies such as WMNs, WiMAX, and power line communication to deploy AMI communication. Miscellaneous AMI routing and scheduling methods are discussed in [10]. This paper represents the network as a graph comprising a set of vertices and edges. Studies [11], [12] indicate that the WMN is suitable to support medium to high range network. It is suitable to implement scalable, cost-effective, and self-healing networks in both rural and urban areas. HWMP is the default routing protocol for wireless mesh networks, drafted by the IEEE 802.11s standard [13]. In [14], the importance of the SINR in the feasible scheduling of the WMN is proposed. Each link is assigned a pre-determined link weight that is used for link activation. The link weight is used to schedule the routes and for routing. Optimising WMNs using routing protocols, such as OLSR and HWMP, with infrastructure architecture and ad-hoc architecture are discussed in [15]. The objective functions of WMN routing are to reduce end-to-end delay and increase throughput. Optimising routing and planning using a genetic algorithm for WMNs is explained in [16]. This paper describes the WMN scenario but does not discuss interference in WMNs.
Jing et al. [17] proposed a new optimisation algorithm called the ACA-SA that finds the optimal paths between smart meters and data collectors in the AMI with load balancing. It results in minimal delay and maximum throughput. The level of congestion is calculated using a load factor throughout the path. Two routing methods are proposed herein. The first is applied for delay-sensitive applications of AMI such as meter reading transmission and outage management, and the second is for AMI applications with low packet loss ratio such as pricing and on-demand response. The proposed GA-LA-HWMP method is compared with the routing method for delay-sensitive AMI applications. Yuan et al. [18] proposed a hybrid routing protocol for WMNs with load balancing. This method divides the load into three levels: a mesh client-level load, a mesh router-level load, and a gateway-level load. The load is balanced within each level, and the results show that the packet loss is high in case of a large number of simultaneous flows.
Sahasrabuddhe and Mukherjee [19] formulated the multicast problem as an optimisation problem with the objective of minimising the average hop distance and number of transceivers. This method reduces the effects of interference. A multi-objective optimisation of WMNs to improve QoS parameters is discussed in [20]. The algorithm contains multiple objectives to enhance the QoS parameters using the ant colony routing method. In [21], AMI is deployed with a WMN that uses Zigbee and IEEE 802.11 technologies. Different routing methods and routing parameters are compared. This paper discusses hybrid network scenarios for AMI with both technologies. In [22], a multipopulation parallel genetic algorithm is used. This paper combines a GA with a simulated annealing technique, and details pertaining to the GA’s application in the deployment of inter-urban mesh networks are explained. In this work, optimisation enhances the routing method to satisfy the standard of IEEE 802.11s for the WMN. In [23], the optimisation of multicast request handling is implemented using a genetic routing algorithm. A number of alternate multicast paths are used to find the optimal path. The objective of the work is to minimise the number of split-capable nodes in the network for a given set of multicast requests. Brar et al. [24] explained the impact of interference in WMN routing. Herein, the communication network is divided into a number of subnets, and the interference within each subnet and between subnets is discussed.
In our previous work [25], the HWMP algorithm was implemented with a load-balancing strategy to avoid congestion at highly demanded smart meters. For a better understanding, the load-balancing algorithm is explained using a case study, as detailed in Section 4 herein. With an increase in the number of smart meters, packets from some smart meters require additional hops. This type of transmission is susceptible to increased delays based on the large number of retransmissions caused by packet drops. The packet drop is caused by the inability to meet the threshold value of the SINR [26] at the receiver. To reduce packet drop, a genetic algorithm is applied in this work to optimise the load-balanced paths. As a result of such optimisation, a fast and reliable transmission is possible with a reduced network load.
The paper is organised as follows: Section 2 describes the AMI architecture with the WMN factors that deploy the AMI network. Section 3 explains the problem formulation for the proposed optimisation. Section 4 discusses the LA-HWMP method with a case study. Section 5 describes the proposed GA optimisation of the LA-HWMP routing protocol. In Section 6, the realisation of the AMI network using the WMN along with GA-LA-HWMP is discussed. In Section 7, the experimental setup of the proposed method is given. The same section analyses different performance metrics under different constraints. Section 8 concludes the paper.
Section snippets
AMI communication
In this work, the AMI is deployed using a WMN where smart meters are represented by mesh routers and AMI data collectors by mesh gateways. A simple AMI architecture with a WMN is shown in Fig. 1. Mesh topology is used in the AMI network because of the correspondence in configuration, and the functionality of the mesh network with AMI [27], as listed below:
- •
In the AMI network, all smart meters cannot access the gateway directly. In such cases, a smart meter can establish a path to the gateway
Problem formulation
Table 1 summarises all the symbols used in this section, along with a description of each.
The AMI communication infrastructure developed using the WMN can be represented as an undirected graph, , where . The set of vertices contains the set of smart meters and the set of data collectors . The graph is undirected and need not be complete, as random mesh network topology is used. The task is to identify a load-balanced path between a meter and a data collector with
Load-Aware HWMP (LA-HWMP)
The IEEE 802.11s uses a HWMP as the default routing protocol. The airtime metric is used to select the nodes for the path, as given in Eq. (8). where O is the overhead latency, is the test frame size, is the data rate in Mb/s, and the frame error rate. The link with the minimal airtime metric is selected to forward the packet. In our previous work [25], a load-balancing strategy was added to the HWMP and named the load-aware HWMP (LA-HWMP). In the WMN, three types of
Genetic algorithm-based optimisation
The genetic algorithm [16] is an optimisation tool that uses a population of solutions, rather than a single solution, to reach the optimal solution. The basic genetic operators are encoding, selection, crossover, and mutation. Encoding is the representation of chromosomes (decision variables) based on the application. The representation can be binary to represent the on–off state of an entity, or an integer to represent a unique value, or a string. Each individual is assigned a fitness value
Genetic algorithm for optimising la-hwmp paths
The HWMP is the recommended routing algorithm for the WMN based on the IEEE 802.11s standard [32]. By default, the HWMP uses two route identification methods, called proactive and reactive modes for static and mobile nodes, respectively. Given that smart meters are static, a proactive routing is used. This method uses the radio metric known as air time metric for path selection. All gateways initialise the route identification process by sending a route announcement (RANN) packet to all mesh
Experimental setup
The proposed GA-based algorithm was implemented using the Proto C language in OPNET modeller 14.5. The process of the GA implementation is given in Fig. 5. The network parameters used to implement the scenarios are given in Table 2.
Four scenarios with 100, 200, 300, and 500 smart meters were simulated to analyse various network parameters such as the end-to-end delay, load, and throughput.
Conclusion
In this paper, the load-balanced paths were optimised using a GA. The routing method was tested for a number of smart meters and acceptable results obtained in line with ANSI standards. The primary objective of minimising the end-to-end transmission delay was achieved by applying the GA on a set of LA-HWMP paths. This was possible by preserving links with a higher SINR value at the receiver to evolve in the next generation. This method ensured scalability, which is a key routing requirement for
References (36)
- et al.
A survey on consumers empowerment, communication technologies, and renewable generation penetration within smart grid
Renew. Sust. Energy Rev.
(2018) - et al.
Smart grid technologies and applications
Renew. Sust. Energy Rev.
(2016) A survey on smart metering and smart grid communication
Renew. Sust. Energy Rev.
(2016)- et al.
Algorithms for smart grid management
Sust. Cities Soc.
(2018) - et al.
A multi population parallel genetic simulated annealing-based QoS routing and wavelength assignment integration algorithm for multicast in optical networks
J. Appl. Soft Comput.
(2009) - et al.
Inter-flow and intra-flow interference mitigation routing in wireless mesh networks
Comput. Netw.
(2017) - et al.
IEEE 80211s wireless mesh networks: framework and challenges
Ad Hoc Netw.
(2008) - et al.
Running time analysis of Ant Colony Optimization for shortest path problems
J. Discrete Algorithms
(2012) - et al.
Smart grid — the new and improved power grid: a survey
IEEE Commun. Surveys Tuts.
(2012) - et al.
Scalable distributed communication architectures to support advanced metering infrastructure in smart grid
IEEE Trans. Parallel Distrib. Syst.
(2012)
A survey on smart grid communication infrastructures: Motivations, requirements and challenges
IEEE Commun. Surveys Tuts.
Routing in neighborhood area networks: a survey in the context of ami communications
J. Netw. Comput. Appl.
IEEE 80211s: The WLAN Mesh Standard
IEEE Wireless Commun.
Cited by (11)
A multi-objective Dyna-Q based routing in wireless mesh network
2021, Applied Soft ComputingCitation Excerpt :Some routing researches focus on a single objective like throughput, delay and packet loss. Load balancing is also essential to be considered to avoid congestion [5], and the network resources like energy can be consumed in balance [6]. However, with the rapidly increasing demands from users, only one objective often fails to improve the whole network performance.
GWO-Based Simulated Annealing Approach for Load Balancing in Cloud for Hosting Container as a Service
2022, Applied Sciences (Switzerland)Energy Efficient Evolutionary Algorithm based Clustering with Route Selection Protocol for IoT Assisted Wireless Sensor Networks
2022, International Journal of Computer Networks and ApplicationsRouting with Ant Colony Optimization in Wireless Mesh Networks
2022, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)