Various techniques for conserving energy in IoT communications have been pro- posed by researchers [29,30]. This section provides an overview of such existing mechanisms. These mechanisms are grouped based on their objectives and key tech- niques for implementation. The analysis of these mechanisms is presented in the next section.
4.1 Mechanisms for Energy-Efficient Communication
The authors of [31] have described an energy-efficient management framework for the IoT environment. The main idea proposed in this paper is to regulate the duty cycles of IoT sensors considering QoI (Quality of Information) requirements. This paper also proposes an idea to cover the critical task set to choose sensor services.
This concept is based on QoI-aware sensor-to-task relevance. The advantage of this proposal is that it can be used with any underlying routing protocols for a variety of applications to preserve energy in communication. Moreover, the energy manage- ment decision is taken dynamically to deal with constraints such as service delay and optimum level of energy. The authors have also considered the latency of process- ing and signal propagation time and shown the impact of these factors on average measured delay probability. The proposed algorithm is greedy in nature and exe- cutes mainly three steps: shutting down sensors which are not perilous to the recent task; keeping the current status of each sensor for every task and computing the least energy requirement likelihood for task transitions. The proposed framework is applicable to certain realistic scenarios.
For energy-efficient and highly scalable IoT, the concept of multihop networking is presented in [32]. The mechanism uses blind cooperation along with multihop communications to improve scalability. Power control is necessary to have efficient blind cooperation. The authors also discuss an uncoordinated power control tech- nique in conjunction with blind cooperative clustering to be implemented in each device. As claimed in the paper, this mechanism outperforms the simple point-to- point routing mechanism. Multihop networking helps reduce the overhead associated with the underlying routing protocols and enhances scalability. An upper bound for the mean transmit power level is computed as a function of cluster size. The proposed mechanism is evaluated based on the normalized transport rate. The performance of this mechanism is improved when there is a small deviation between the real size and assessed the size of a cooperative cluster.
According to the authors of [33], Ferroelectric RAM (FRAM) technology can be used in IoT edge devices to control unreliable power supply. FRAM is a volatile memory technology and can be treated as a unified memory. FRAM-based solu- tions offer reliability but they are energy inefficient in comparison of SRAM due to a greater access latency. In contrast to FRAM-based solutions, SRAM-based solu- tions provide a high level of energy efficiency but they are not reliable in the face of power loss. The authors present a hybrid approach which is based on FRAM- SRAM MCUs and uses shrewd memory mapping to get reliability from FRAM and efficiency of SRAM-based systems. The memory mapping technique proposed in this paper is energy-aware and lowers the consumption of energy without affecting reliability. The proposed technique uses eM-map-based representation to compute the optimum memory map for the functions which establish programs. This makes the solution platform-portable and energy-aligned. To enhance energy efficiency and performance, the solution aligns system’s powered-on time intervals to function exe- cution bounds. The authors claim a 20% reduction in energy consumption when their proposed solution is used.
Van et al. [34] propose the use of converged Fi-Wi (Fiber-Wireless) access net- works to develop a collective communication facility for the Internet of Things.
The paper discusses the applicability and difficulties associated with the design and implementation of energy-efficient IoT infrastructures in the optical backhaul net- work. This effort discusses the usage of converged Fi-Wi networks which combine a capacity-centric OAN backhaul and a coverage-centric multi-RAT front-end net- work to support the Internet of Things infrastructure. The paper proposes mecha- nisms for power saving, scalability, energy efficiency, H2H/M2M coexistence and network integration for IoT deployment scenarios. The Authors state that in small- scale Fi-Wi-based IoT scenarios, approximately 95% of energy can be preserved by implementing TDMA-based scheduling. In the large-scale LTE-based IoT setups, up to 5 years of battery life can be attained by incorporating the suggested DRX technique.
Suresh et al. [35] describe an energy-efficient mechanism called EEIoT (Energy- Efficient Internet of Things) for IoT applications. This mechanism is based on the concept of MECA (Minimum Energy Consumption Algorithm). MECA techniques are not efficient as they do not consider energy consumption in sensor nodes. The goal of EEIoT is to control factors that energy consumption in IoT efficiently. EEIoT has a self-adaptation property and can lower energy harvesting to a great extent in the IoT environment. Different energy consumption factors have been considered by the authors to make the implementation of EEIoT effective. The paper also describes the effective method of dealing with energy efficiency requirements for data streams in big data platforms. Here, the objective is to maintain connectivity and normal functionality despite low energy levels. According to the authors, EEIoT outperforms other traditional methods of energy saving.
Energy Conservation in Multimedia Big Data Computing … 47
4.2 Mechanisms for Energy-Efficient Routing
A hybrid routing protocol for M2M sensor networks for wireless IoT applications has been proposed in [36]. In this paper, nonuniform energy consumption, scalability of network and performance degradation issues have been addressed. To handle these issues, a scalable energy-efficient clustering method is presented. In this approach, numerous mobile sink nodes have been considered for large-scale M2M sensor net- works to discover the shortest routing trail to the destination node from the cluster head. It helps to lengthen the lifespan of the network. The rotation used in the selec- tion of cluster heads results in a better distribution of energy and traffic among all sensing nodes in the network. Because of this, the lifetime of the nodes and hence of the network is increased. It also helps in improving end-to-end delay and through- put of the communication process. Simulation outcomes indicate that the use of the suggested clustering procedure and multiple mobile sink nodes enhance the network lifetime and energy distribution. The authors claim significant improvement in the routing process compared to the efficiency of the existing routing protocol.
Alkuhlani and Thorat [37] address a trade-off between the privacy of locations and energy consumption in the IoT environment. The authors of this paper suggest a secure procedure to preserve the location secrecy of the source node by modifying the routing procedure, and energy-aware load balancing protocol based on the selection of random routes to support fair energy consumption with path diversity. In this method, each packet is forwarded to a random node. To keep the actual route of the packet confidential and maximize the confidentiality, tunnels of M intermediate hops are defined. This keeps the location of the source node hidden from the attacks based on backtracking. The paper explains an energy-aware load balancing RPL-based protocol which is integrated with multiple random path-finding techniques such that each packet is forwarded in different directions on a random basis. This idea prevents attackers to eavesdrop the exact location of the source node using back-tracing and may reach a different node.
An energy proficient self-organizing multicast routing protocol called ESMR is presented in [38] for IoT applications. In this protocol, there are two categories of nodes: network nodes and nonnetwork nodes. The nodes which are present in the network are treated as network nodes. The other category of nodes, nonnetwork nodes use Markov process based diverse measurements for computing a network nodes’ weight. The protocol selects a node with the highest weight as a sink node.
Nonnetwork nodes can enter into the network by sending requests to sink nodes.
This is how a tree-like structure is constructed in the network. The structure of the network is balanced in stages to control energy levels. Automatic AVL tree pruning operation is used to prolong the network lifetime. When the network grows in size, the packet loss proportion does not increase due to pruning and this results in extending the lifespan of the network. In the AVL tree structure, a node is declared as a sink node only if can poise child nodes, remaining energy, hop count and the spatial deviation between the sink node and its child nodes. The height of the tree is also optimized in ESMR. The topology of the network is changed during data
communication based on the future energy of sink nodes. The authors of this paper claim that a dependable tree-based network can be built using ESMR which improves the lifetime of the network and lowers the consumption of energy. The success rate of packets is shown to be improved in ESMR compared to AODV (Ad hoc On-Demand Distance Vector Routing), DSDV (Destination Sequence Distance Vector Routing), and ADMR (Adaptive Demand-Driven Multicast Routing) protocols.
Under many application areas in the IoT, the mobility of nodes and P2P (Pointto- Point) communication are the basic requirements. Therefore, such applications should have routing mechanisms that support mobility and discovery of best P2P paths and focus on energy efficiency as well. According to the authors of [39], the existing P2P routing protocols for the IoT do not upkeep the movement of nodes.
Hence, they suggest a novel energy-efficient mobility-aware routing protocol titled MAEER (Mobility-Aware Energy-Efficient Routing) for the IoT scenarios. This pro- tocol minimizes the total number of partaking nodes in the P2P path-finding process to lower the ingesting of energy. It also provides a mechanism to facilitate mobility with improved packet delivery ratio. As stated in this paper, the energy depletion of MAEER is 24% less compared to P2P-RPL protocol.
Behera et al. [40] describe the altered LEACH (Low Energy Adaptive Clustering Hierarchy) protocol to decrease the energy ingesting in sensor nodes for IoT appli- cations. This protocol defines a threshold for selecting cluster heads with simultane- ously changing the power levels between the nodes. The authors state that their mod- ified LEACH protocol performs better than the existing LEACH and other energy- efficient protocols in terms of network lifespan, throughput, and stability period when used in different scenarios with varying network size, the density of nodes and available energy.
4.3 Mechanisms for Self-generating and Recycling Energy/Green IoT
Shaikh et al. [41] describe the efficient deployment of different technologies such as sensors, Internet, and other smart objects in the IoT environment to make it green IoT. According to the definition given in this paper, Green IoT is defined as the IoT which uses either hardware or software-based energy-efficient procedures. The objective of green IoT is to diminish the effect of the greenhouse effect of IoT services and applications. The life sequence of green IoT mainly focuses on reducing the greenhouse effect in the design, production, utilization, and disposal or cycling processes. The authors of these paper have also considered numerous facets such as applications, communications, crucial enablers, and services to achieve green IoT.
The survey of various solutions for achieving green IoT is also presented. The paper provides the list of IoT application areas where it is possible to conserve energy to have the green environment. The list of key enablers of green IoT and various methods to implement energy-efficient solutions with respect to these enablers are
Energy Conservation in Multimedia Big Data Computing … 49 also discussed. Domains of green IoT such as service management, heterogeneous communication, physical environments, and sensor cloud integration are required to be considered for providing efficient communication among them. Future scope in existing efforts is highlighted to implement green IoT.
An energy-efficient protocol stack named GREENNET is presented in [42]. This solution is proposed for IP-enabled wireless sensor networks but can be used in the IoT environment. The protocol stack executes on a photovoltaic cell energy- enabled hardware platform. GREENNET integrates different standard mechanisms and improves the performance of existing protocols to a great extent. It provides a discovery mechanism that facilitates the adjustment of the duty cycles of harvested nodes to the remaining energy in the network and leverages network performance.
GREENNET also supports the security of standard operations at the link layer and data payload. It does not utilize multiple channels to increase the capacity of the net- work. Robustness and mobility of the nodes are considered in the proposed scheme.
An amended tiered clustering protocol named EH-mulSEP (Energy Harvesting enabled Multi-level Stable Election Protocol) is given in [43] for green IoT-based het- erogeneous wireless sensor networks. The authors of this paper discuss the effects of energy harvesting methods in large-scale IoT systems when a large number of relay nodes that harvest energy and obtain the accumulated information from the selected cluster heads. Relay nodes forward this information to the base stations. The paper also presents a general computation method for multi-level weighted election possibility which can facilitate up to n levels of heterogeneous nodes with their cor- responding level of primary energies. More than three types of nodes are considered for heterogeneity, which provides generic models for higher initial energy levels in sensor nodes. The major goal of EH-mulSEP is to reduce the energy depletion in battery-operated sensor nodes in IoT applications and maximize the conservation of energy by increasing the scalability and lifespan of the network. Using an intermedi- ary energy harvesting layer between the base stations and cluster heads, EH-mulSEP improves the performance of the network in terms of throughput, permanence, scala- bility, network lifespan, and energy consumption in comparison of the other versions of SEP protocols in similar deployment settings.
4.4 Mechanisms for Storing and Harvesting Energy
IoT systems need sensing, data congregation, storage, handling, and data transmis- sion capabilities. Real time, as well as virtual sensors, are used to provide these capabilities. Mahapatra et al. [44] describe robustness in data delivery process and energy efficiency as the major requirements of IoT communication. The authors of [44] propose data awareness, cluster head selection using active RFID tags and energy harvesting in the IoT the environment. The proposed protocol, DAEECI (Data Aware Energy-Efficient Distributed Clustering protocol for IoT), saves energy involved in the cluster head selection process. It uses active RFID tags to reduce dispensation energy by including data awareness factor and enhancing lifespan by infusing RF
energy harvesting. Energy consumption models are formulated in each round and the same is sent from sensor nodes to BS through gateways. The authors claim a significant enhancement in network lifetime and data delivery when DAEECI is deployed.
A novel sensor architecture named EcoSense is presented in [45]. Unlike conven- tional software-based techniques, EcoSense uses a hardware-based reactive sensing technique that removes the energy waste generated by a sensor working in either standby mode or sleep mode. If the target events are available, a sensor is powered off to reserve energy. When the target events are present, a reactive connection com- ponent harvests energy from the events and activates the sensor again. Light and RF-driven sensors are used to sense lights and RF signals and provide fair reac- tion distances. The reactive connection module is used to control the connections between the power supply unit and sensors. By default, this module is disabled so no power is used by the sensors. When a target event takes place, the energy harvester module stores the energy correlated to the events and enables sensors by linking them to the power supply. The performance is evaluated based on reaction distance, reaction times, and working duration. Results state that the suggested mechanism is applicable only in short-range applications.
A modified routing mechanism for the 802.11 networks is presented in [46]. The routing protocol proposed in this paper uses energy harvesting data for making path- finding decisions. The objective of this modified routing protocol is to extend the network lifetime when it is set up in an energy-constrained scenario. When no viable energy source is available, network nodes harvest energy from the environment. This logic is incorporated into the routing activity so that the network operation can exe- cute without disruptions and harvested energy can be utilized properly. The routing algorithm determines energy-efficient routes for transmitting messages through the network. To achieve this, the algorithm maintains energy harvesting information along with the level of residual energy at each hop in the network. To see the effect of the proposed routing protocol on energy management, the authors presented the simulation of the proposed logic with varying energy harvesting conditions. Accord- ing to the results given in the paper, the proposed protocol can improve network lifetime by approximately 30% in low energy harvested scenarios. In high energy harvested conditions, the proposed algorithm is claimed to avert the energy hole problem successfully.
The infrastructure of the IoT comprises of a big number of battery-driven devices with restricted lifetime. The manual replacement of their batteries is not feasible in large-scale deployments. The host stations need to communicate with the distributed sensor devices and this communication requires a significant amount of energy based on the physical distance between the host and sensing nodes. An energy-efficient multi-sensing platform is presented in [47]. This paper addresses long-range device communication, energy harvesting and self-sustainability of low-power short-range devices in the network. The idea to design a power-efficient solution and reduce the quiescent current in the radio devices even when they are always on in the wireless channel. The proposed platform supports a heterogeneous long and short- range network architecture to minimize latency and energy consumption during the
Energy Conservation in Multimedia Big Data Computing … 51 listening phase. For better energy management, the architecture combines LoRaTM and wake up radio. This results in increased communication efficiency and reduced power consumption.
Yang et al. [48] explore resource distribution for a machine to machine-aided cellular network to achieve energy efficiency and nonlinear energy harvesting. The proposed method uses two major access strategies, NOMA (Non-Orthogonal Multi- ple Access) and TDMA (Time-Division Multiple Access). This method attempts to reduce the total energy consumption in the network through joint circuit power con- trol and time allocation. The authors state that both access strategies can be used for optimal machine communication with minimum energy consumption and improved throughput. Energy consumption of each machine type communication device is defined as a convex function with regard to the assigned communication duration.
Using the optimum transmission power conditions of machine type communication devices, the optimization issue for NOMA can be transformed into an equivalent issue whose solution can be derived suboptimally. The paper also discusses the transfor- mation of the original TDMA optimization to an equivalent tractable problem by considering appropriate variable transformation. This transformed problem can then be solved iteratively. The authors show that NOMA requires less amount of energy compared to TDMA with low circuit power control machine type communication devices. In the case of high circuit power control of machine type communica- tion devices, TDMA does better than NOMA, in terms of energy efficiency. The paper also analyses the total energy consumed in NOMA and TDMA policies in uplink M2M communications. Energy minimization problem is stated in terms of circuit power consumption, throughput, energy causality, and transmission power constraints. Either NOMA or TDMA can be used based on the circuit power control in machine type communication devices.