Secured Protocol with Collaborative IoT-Enabled Sustainable Communication Using Artificial Intelligence Technique

Abstract

In recent years, 5G and the Internet of Things (IoT) have been integrated into a variety of applications to support sustainable communication systems. In the presence of intermediate hardware, IoT devices collect the network data and transfer them to cloud technologies. The interconnect machines provide essential information to the connected devices over the Internet. Many solutions have been proposed to address the dynamic and unexpected characteristics of IoT-based networks and to support smart developments. However, more work needs to explore efficient quality-aware data routing for distributed processing. Additionally, to handle the massive amount of data created by smart cities and achieve the transportation objectives for resource restrictions, artificial intelligence (AI)-oriented approaches are necessary. This research proposes a secured protocol with collaborative learning for IoT-enabled sustainable communication using AI techniques. This approach increases systems’ reaction times in critical conditions and also controls the smart functionalities for inter-device communication. Furthermore, fitness computing can help in balancing the contribution of quality-aware metrics to achieve load balancing and efficient energy consumption. To deal with security, IoT communication is broken down into stages, resulting in a more dependable network for unpredictable environments. The simulation results of the proposed protocol have been compared to existing approaches and improved the performance of response time by 17%, energy consumption by 14%, number of re-transmissions by 16%, and computing overhead by 16%, under a varying number of nodes and data packets.

1. Introduction

The concept of IoT has gained the interest of the academic community, with the ultimate goal of connecting sensors, actuators and smart appliances to a common interface that allows them to communicate with one another [1,2,3]. It has grown in terms of both the number of users and the number of available smart devices, which together can link to create smart cities. Sensors have become a part of our daily lives in a variety of industries as a result of rapid improvements in wireless communications and digital electronics. Wireless Sensor Networks (WSNs), which are made up of a variety of small devices, have emerged as a critical technology for smart systems and interact with IoT networks [4,5,6]. The sensor-based revolution of the WSN-based IoT, in particular, has resulted in significant technical advancements in practically every aspect of our lives [7,8,9]. IoT-based sensor networks are used in smart cities to monitor and regulate physical quantities such as temperature, humidity, pressure, acceleration, and so on. The primary function of these sensor nodes is to gather data with heterogeneous resources and send them to cellular stations [10,11,12]. A vast number of observations from various sensor nodes must be collected and integrated to provide a thorough description of an environment or to reach a solid conclusion. It can be noted that only a small percentage of the data in these massive databases is useful, while the rest is meaningless [13,14,15]. Constraint IoT devices have become increasingly ubiquitous in recent years, resulting in huge amounts of data being created in real-time, which makes AI systems an interesting target. However, managing such large amounts of data is a difficult process. Machine learning models, on the other hand, are challenging to implement on end devices. A common technique is to offload data to other computing systems for additional processing; however, doing so increases latency and connection costs [16,17,18]. To overcome this problem, many efforts have been made to locate more computing power near the network’s edge and support the IoT system with the least overhead. Data routing has recently become an important approach for any IoT system to select a route, and numerous solutions have been researched [19,20,21]. On the other hand, few systems offer security analysis on the existence of internal and external network threats. Many cryptography-based algorithms have been explored to validate the security of IoT network solutions [22,23]. Various attacks on the algorithms in the presence of faulty or malicious nodes have been tested. However, some of them can impose low communication overheads and increase the efficacy of the network system. In order to provide efficient routing services for smart devices in terms of response time, resource utilization, and latency, this study proposes a secured collaborative intelligence protocol for sustainable networks. Moreover, the proposed protocol further protects the confidentiality of network systems by foreseeing connection instability in an unreliable environment.

The following are the contributions made in this research:

i.

In the distributed processing of IoT systems, the proposed protocol provides a collaborative intelligence for the multi-path selection technique. Additionally, it offers more effective distributed services for networked devices and increases resource utilization.

ii.

The collaborative information is updated using a multi-parametric function that balances the contributions of each metric and improves the reception of IoT data over the inherently unreliable network.

iii.

Furthermore, to strengthen the trustworthiness of sustainable systems, security methods are applied to multi-stages.

iv.

Extensive tests demonstrate the significant improvement of the proposed protocol for routing and security analysis.

The remainder of the research is organized as follows: In Section 2, the related work and the formulation of the research challenges are highlighted. The significance of the suggested protocol and its components is examined in Section 3. Section 4 explains the experimental results as well as their comparison to existing solutions. Section 5 ends with suggestions for future research.

2. Related Work

In future Internet technologies, IoT-oriented constraint networks provide novel connections between physical and cybernetic components. As such systems become more widely utilized, they need to improve their computational analysis and complexity [24,25,26]. IoT systems have become an integral part of most modern businesses and industries; however, due to the vast amount of heterogeneous data incoming from multiple sensing devices, they face many issues, including long-term performance, limited throughput, and data security [27,28,29]. A great deal of research has gone into reducing the energy consumption of WSNs by improving data reception. Designing a hierarchical clustering method is one of the many techniques for decreasing the energy consumption of IoT-oriented systems [30,31]. Each cluster has a single special node called a cluster head for data aggregation and the forwarding of the cluster’s data. In such schemes, communication can be intra-cluster and inter-cluster. Intra-cluster communication is defined as communication that occurs within the cluster region. However, data transmission between clusters is referred to as inter-cluster communication. In [32], the conventional low-energy adaptive clustering hierarchy (LEACH) methodology was improved and included a threshold limit for cluster head selection. Moreover, it managed the power level among nodes at the same time and increased the efficacy of delivery. Authors in [33] proposed the energy-harvesting-aware routing algorithm (EHARA), which was improved by a new parameter called “energy back-off.” The proposed approach enhances the lifetime of nodes and the network’s quality-of-service (QoS) under varying traffic loads and energy availability scenarios by combining multiple energy harvesting strategies. Energy harvesting is the process of obtaining energy by exploring the environment and other sources. Such technology delivers sustainable networks and energy-efficient data routing techniques. Data latency and transmission errors are decreased by implementing QoS over the network communication system. Additionally, QoS metrics effectively study the communication bandwidth and overheads. In [34] authors investigated an energy-efficient clustering technique for multiple-input multiple-output (MIMO) IoT communication systems. For IoT applications in the 5G environment and beyond, a novel MIMO-based energy-efficient unequal hybrid clustering (MIMO-HC) protocol is suggested. The efficiency of the proposed MIMO-HC protocol was evaluated using experimental methods and compared to existing state-of-the-art research. In comparison to existing approaches, the suggested MIMO-HC scheme consumes less energy and has a longer network lifetime. In [35], the authors presented a reinforcement learning technique to address concerns with multi-hop data transmission, such as increased data latency, higher interference, and lower data throughput (i.e., poor bandwidth use). The proposed method updates the network’s Q-matrix at periodic intervals and selects relay devices to maximize the cumulative reward value. When compared to other existing methodologies, the results reveal that the proposed strategy improves network performance in terms of energy efficiency and quality of service. Authors [36] introduced “CTrust-RPL,” a unique hierarchical trust-based method that assesses node trust based on their forwarding actions. To preserve computational, storage, and energy resources at the node level, this study sends difficult trust-related computations to a higher layer known as the controller. It also compares the proposed mechanism to Sec-trust, a state-of-the-art technique. The proposed system outperforms the competition in terms of identifying and isolating blackhole assaults. In accordance with the related studies, it is found the energy efficiency is still a burning research issue for sustainable applications. It is seen that most of the proposed works offer the efficient utilization of energy, but at the cost of overheads on nodes. Moreover, most of the proposed solutions do not consider the realistic parameters for intelligent system decisions and as a result increase the number of re-transmissions and the network cost. Although several solutions have been proposed to cope with security measurement for constraint IoT applications, they impose additional burdens for the communication links.

Problem Identification

Numerous hardware-oriented communication methods have been identified in the studies of IoT-based computing systems. These technologies enable the efficient delivery of critical information to faraway remote users over the 5G or 6G mobile networks. However, high restrictions on wireless devices incur the issue of energy efficiency and balanced traffic distribution. Moreover, a few solutions have been proposed using the intelligence of collaborative learning, but such approaches do not explore the network circumstances for a robust system decision. In addition, most of the techniques in IoT systems are not able to cope with internal and external network privacy attacks and attain sustainable applications. They overlook the smart gateway communication with the support of trust in an unpredictable environment. Based on the above-mentioned limitations, providing an energy-efficient solution is a demanding task within the integration of AI techniques. Such a system not only gives a balanced transmission rate but also effectively controls the resources of constraint nodes.

3. Proposed Secured Protocol with Collaborative Intelligence

In this section, we first give a brief overview of the proposed protocol, followed by a discussion of its developed phases. The proposed protocol is comprised of numerous small and less-expansive sensor nodes. They can communicate with each other by exploring their transmission distance. All the nodes have limited energy, transmission, and processing power. Network data can be transmitted from the source node to the sink node by exploring gateway nodes. Nodes cannot be charged from any external source. Each gateway must determine the trustworthiness of incoming data as well as authenticate the sensors. The sensors must establish trust with the gateway nodes using secret keys, and only trusted data are passed to the mobile sink for further analysis and processing. The mobile sink is rotated in the clockwise direction to lower the routing holes and increase the energy efficiency of the network structure. To create the network structure, nodes are arranged in the form of a unidirectional graph, $G$, with edges, $E$, and nodes, $N$.

3.1. Routing Phase with Collaborative Intelligence

This section describes the process of data routing from source to mobile sink with the support of the collaborative information sharing of devices. To achieve this, each node must compute its heuristics value and share the information with its localized neighbors. Based on the collected information, each node easily selects the more optimal node as a forwarder. For node $x$, the proposed protocol employs the A* algorithm [37,38] with a heuristic function $f\left(x\right)$, which is made up of forwarding cost, $h\left(x\right)$, and backward cost, $g\left(x\right)$, as shown below.

$$f\left(x\right)=h\left(x\right)+g\left(x\right)$$

(1)

$h\left(x\right)$measures the cost of the source node $x$ from the goal node, whereas $g\left(x\right)$ denotes the cumulative cost of a node $x$ from the root node. In the proposed protocol, the heuristic function is computed based on the multi-parameters i.e., energy, $e$, processing delay, $p$, and available memory space, $ms$, as given by,

$$\alpha =\left(e, p, ms\right)$$

(2)

Subject to:

$$e={\sum }_{i=0}^n(t_i+r_i)$$

(3)

$$p=|r_t-s_t|/P$$

(4)

In Equation (3), $t_i$is transmitted and $r_i$ is received packets. Equation (4) defines the processing delay in terms of time for packets sending $s_t$ and packets’ reception $r_t,$ $P$ is the number of packets. To compute the memory usage $X$, consumed space $C_s$, buffer size $b_s$, and memory size $M$ are explored, as given by,

$$X=\left(C_s+b_s\right)/M$$

(5)

$$ms=M-X$$

(6)

The proposed protocol first identifies the set of routes based on a forwarding cost $h\left(x\right).$ Such routes are established from the source node towards the goal node. Let us consider that nodes are denoted by $N=n_1, n_2, n_3, \dots , n_k,$ then the computed forwarding cost $R$ for an individual node is given by,

$$R={\alpha }_{n1 }, {\alpha }_{n2 }, {\alpha }_{n3 }, \dots , {\alpha }_{nk }$$

(7)

Afterwards, the proposed protocol selects the optimized route from the set of routes as defined below.

$$g\left(x\right)={\sum }_{i=0}^n{g}_i\left(x\right)$$

(8)

Figure 1 shows the proposed methodology for choosing the best path by examining shared intelligence. The main stage is the computation of heuristic values that must be minimized. There are two values; one is for direct measurement and the other is for exploring by intermediate nodes. The forward cost is used to compute the value for a source node to a gateway node in a direct transmission link, whereas the backward cost is used to identify the summation of intermediate costs from the source node to the gateway.

3.2. Security Phase with Authentication and Trustworthiness

In the next phase, the proposed protocol provides security in terms of authentication and privacy by exploring cryptography techniques. In the presence of malicious identities, gateway nodes are essential for authenticating the trusted nodes and preventing network attacks. In the proposed security phase, the malicious nodes are identified to overcome the privacy and integrity issues. Malicious nodes send false information to divert the sensors’ data on compromised routes and ultimately leak the confidentiality of data. Such nodes observe the communicated data and also destruct the wireless channels. In the local tables, gateway nodes must first have the distinct identities, $id$, of the normal nodes. The gateway node then creates a secret key, $k$, and securely unicast it to particular nodes. Both the identities and secret keys are grouped in the formation of trust value. After verifying the trust, the node is declared authentic and is allowed to forward the collected data toward the mobile sink. To begin the trust process, the source node, $x,$integrates both the $id$ and $k$ values, and later an encryption function is applied to the data, $D$, with the utilization of $k$, as given below.

$$Y=id+k$$

(9)

$$D^{\prime }=D+k$$

(10)

As a result, all the information is grouped in a single packet and transmitted to gateway nodes, as given by,

$$Z_{x, G}=\left(D^{\prime }, id, k\right)$$

(11)

When the gateway node receives the data, it extracts the $id$, $k$, and $D^{\prime }$ information. Along with data decryption, it uses the xor function to verify the requestee node’s identity, as defined in Equations (12) and (13).

$$xor\left(id, k\right)=Y$$

(12)

$$xor(D^{\prime }, k)=D$$

(13)

The block diagram of the proposed protocol is presented in Figure 2. It emphasizes the developed components as well as their association. There are two sub-blocks; one is for initial network creation with route formulation using the artificial intelligence technique, followed by a more optimal decision-making process for data forwarding by exploring the multi-parameter heuristics. Furthermore, a trusted computation is performed with the integration of node identities and secret keys. The gateway node first has to verify the authenticity of the requestee node by recomputation on the received information and if it obtains the actual trust value, then the node is declared as authentic. Later, gathered data can be transmitted to the sink node with the intelligence of gateways.

4. Performance Evaluation

This section provides the experimental environment, as well as a discussion of the outcomes. Simulations are used to examine and verify the proposed model for IoT-based sensors that are randomly dispersed in the 300 m × 300 m dimension. The sensors’ initial energy is set at 2 J. The range of their transmission is set at 5 m. Mobile sinks revolve at a constant speed around the field’s border. The number of gateways varies, but they all have the resources to handle and analyze IoT data. Malicious nodes are also deployed randomly in the simulation environment to test the proposed model’s security in the presence of false and spurious route formulation packets. Asymmetric communication links exist. The size of the packet has been set to 32 bytes. The number of sensors might range from 100 to 500. We evaluated the security of the proposed protocol against various attacks using cryptography algorithms. We primarily tested the proposed system for privacy, trust, and authentication. Additionally, due to the security of the proposed protocol, performance factors were efficiently utilized.

Table 1. Security work with performance efficacy for the proposed protocol.

Attack	Proposed Procedure
Malicious packets	Nodes’ unique identities
Compromised nodes	Computation of trust value
Privacy	Data encryption using lightweight xor operations
Authentication	Gateway extracts the node’s identity and compares it with the actual value
Untrusted nodes	Verification based on the unique identity and shared secret key
Data integrity	Data hashing

depicts the evaluation of network threats for the proposed protocol.

Moreover, simulation-based configurations are listed in

Table 2. Simulation parameters.

Parameter	Value
Number of nodes	100, 200, 300, 400, 500
Deployment	Random
Sink node	3
Network dimension	300 m × 300 m
Transmission range	3 m
Packet size	64 bytes
Initial energy	2 J
Time intervals	1500 s
Simulations run	15
Gateways	2–10
Malicious devices	10

. Network parameters such as energy consumption, response time, number of re-transmissions, and computing overheads are used to conduct the experiments under a varying number of nodes and a varying number of packets.

We evaluated the performance of the proposed model against the existing solution in terms of energy consumption. Figure 3a,b illustrates the performance evaluation and it is observed that the proposed model improves the efficiency of the energy utilization of the proposed model by 14% and 17%. It was seen that by increasing the number of nodes, the rate of energy consumption is also increased. However, the proposed model offers an intelligent energy solution using a forward and backward cost, and as a result, it selects the updated routes based on certain conditions. Due to fewer control messages and less retransmission, it also balances the energy consumption of the smart communication system and increases the network lifetime. Moreover, routing states are only updated whenever the network edges realize the uncertain situation of the network. Accordingly, the proposed model supports the efficient power distribution system for smart communication with manageable communication costs.

In terms of response time, we compared the performance of the proposed model with the existing solutions. The proposed model greatly reduces the response time, by 12% and 17%, respectively, as shown in Figure 4a,b. This is because gateway devices are integrated across the intended area’s boundary and they are able to handle the rapid transmission of sensor data to the mobile sinks. In addition, by removing faulty and untrusted nodes from routing states, the mobile sink reduces the time it takes to collect data from the IoT system. Additionally, the security solution reduces unwanted traffic on the open transmission network and stops malicious devices from sending false route request packets. As a result, with a nominal delay rate, the proposed model improves the response time for the most critical operations.

In the presence of malicious nodes, Figure 5a,b shows the performance evaluation of the number of re-transmissions under a various number of nodes and data packets. It was demonstrated that by exploring the cryptography-based secured approaches, the proposed model lowers the number of retransmissions by 13% and 15%. This is because the devices’ mutual authentication uses random numbers, unique identities, and secret key generation. The secret keys and random numbers improve the IoT system’s data availability and the rapid identification of malicious devices. Furthermore, gateway nodes are more robust and operate as a supervisor for sensor data arriving from constrained devices; following proper authentication, the obtained data is delivered to application users with the help of edge computing.

The performance comparison of the proposed model against the existing solutions in terms of computing overhead is shown in Figure 6a,b. The proposed model reduces communication load by 12% and 16% for various normal nodes and varying data packets, respectively. This is due to its employment of A* heuristics for learning the communication process and efficiently regulating the power distribution among sensors using a fitness cost. The proposed model re-evaluates routing states whenever any unreliable links are found by packet reception probabilities. As a result, the routing statuses are updated using artificial intelligence techniques, by pre-determined criteria. In addition to lowering sensor overhead, the IoT communication system is managed with effective optimality by lightweight computing services.

5. Conclusions

IoT-based systems offer remote data monitoring and aggregation from a variety of sources and perform different processing phases in order to analyze big data. This study describes a secure and intelligent learning protocol for a sustainable system using AI techniques, unlike most of the existing solutions that do not consider the parameters of quality assurance for distributing services over an unreliable network, and impose extra overheads on resources for attaining trustworthiness. With the assistance of IoT devices, the proposed protocol offers a high-performance computing technique with lightweight route formulation processes. Moreover, the proposed methodology uses artificial intelligence to optimize the backward and forward routing cost to maintain stable connections among devices and improved data delivery performance. Furthermore, gateways give smart systems authentic features by analyzing trust with secret keys and unique identities. Extensive testing was conducted, and the findings showed that the proposed protocol outperformed previous studies. In future work, we intend to train the proposed protocol for distributed services with security assaults using deep learning techniques.