In recent years, with the advancement of the automotive industry and the innovation of electronic technology, issues related to the field of Internet of Vehicles (IoV for short) have received increasing attention. Existing communication technologies can no longer meet the demands of high mobility, low latency, high reliability, and massive streaming data distribution faced by IoV in the future. As one of the key technologies of 5G, D2D communication technology can effectively alleviate the above challenges faced by IoV. This technology not only allows terminals to perform point-to-point direct communication within a certain range but also allows data forwarding between receiving terminals, thereby improving the data distribution efficiency of the IoV, which injects new impetus into the development of the IoV business in the future. However, existing cooperative distribution algorithms between terminals do not fully consider the differences in D2D links, making it difficult to achieve efficient use of spectrum resources. To address this problem, this paper proposes a cooperative data distribution algorithm based on multihop relay in the IoV environment where D2D is used to realize flexible terminal communication between vehicles, including both multicast and unicast modes. The algorithm adaptively selects the optimal relay, route, and number of transmission hops based on the D2D link quality to find the optimal cooperative distribution scheme to achieve the goal of optimizing the energy efficiency of the requesting vehicle. Simulation results show that, compared with the existing one-hop unicast and one-hop multicast forwarding algorithms, the multihop relay data cooperative distribution algorithm proposed in this paper exhibits strong global optimization ability and can significantly improve D2D forwarding resource utilization and data transmission rate and improve the throughput of data distribution services, thereby effectively improving the energy efficiency of the requesting vehicle.

1. Introduction

As living standards continue to improve, more and more people are choosing cars as a means of transportation when traveling. Although cars bring convenience to people’s lives, traffic accidents and traffic congestion are becoming more and more serious as the number of cars continues to grow, causing many problems to people’s daily travel. On the other hand, with the widespread popularization and application of 5G technology and diverse in-vehicle devices, people’s demand for in-vehicle information services has also increased significantly, driving the development of the automotive industry towards the trend of networking and intelligence. In this context, an intelligent information network service based on Mobile Ad hoc Networks (MANETs), that is, Internet of Vehicles (collectively referred to as IoV in subsequent introductions), has emerged and is gradually becoming a research hotspot in the automotive and communication industries [1]. As one of the key elements of intelligent transport system (ITS), the IoV uses advanced mobile communication technology and artificial intelligence technologies to realize the interconnection between vehicles to everything (V2X) so as to help car users obtain real-time access to road live, positioning and navigation, safety warning, in-vehicle entertainment, and other business services [2].

Currently, the typical technologies in the field of IoV are dedicated short-range communication (DSRC) [3] and the physical layer standard 802.11p [4]. However, when these two technologies are applied to some communication scenarios in IoV, such as spectrum shortage or high-density vehicle scenarios in large cities, data packet collisions can occur. This will block the communication link to a certain extent, thereby greatly reducing the overall communication performance of the IoV. To solve this problem, the cellular vehicle to everything (C-V2X) technology based on cellular mobile communication systems has been well received [5]. However, current cellular mobile communication systems are usually based on a predefined infrastructure. This means that point-to-point communication between terminals located in this system is not possible. When two terminals need to interact with each other, they have to rely on the involvement of a wired backbone or a central control node for data forwarding. As a result, when there are more vehicles in the same cluster of the IoV, the channel quality of communication is not guaranteed, thus limiting the system capacity. Therefore, it is essential to explore a new communication technology to improve the communication performance of the IoV. D2D communication is a novel direct communication technology between terminals applied in cellular systems. It allows mobile terminals to use cellular systems licensed frequency bands for point-to-point communications under the control of cellular systems [6, 7]. Since D2D has significant advantages in reducing terminal transmit power, extending battery life, increasing unicast, multicast, and broadcast transmission rates, and improving spectrum efficiency of the system, it has gradually become a new hotspot in the field of wireless communications. Meanwhile, cooperative data distribution is one of the research hotspots. Data dissemination in cellular networks means that some fixed nodes in the network with specific data use unicast, multicast, or broadcast to send data to multiple mobile terminals through wireless links so as to provide users with specified information services [8].

Data distribution techniques in IoV usually rely on V2V or V2I transmission to satisfy vehicle requests by means of one-hop or multihop routing. Based on relevant literature studies, we can summarize the existing data distribution techniques in the IoV into two main cases. The first case is based on data distribution in the IoV without infrastructure. These algorithms are often able to achieve data interaction between vehicles in the IoV through one-hop or multihop communication and achieve the purpose of reducing latency. Moreover, the research focus of these algorithms is relatively more on issues such as hop count [9], transmission capacity [10], and vehicle safety [11], and some of the representative algorithmic models have been widely used in the automotive industry, especially in safety-oriented services. To address the problem of disconnection in IoV, a new mechanism of store-carry-forward is proposed in literature [12]. The core of this mechanism is its use of the geographical area of interest as a reference for dealing with broadcast storms. Literature [13] studies the problem of cooperative data distribution, that is, constructing a path from a source vehicle to a destination through cooperation between vehicles to improve the success rate and rate of data packet transmission. Literature [14] studies the data distribution algorithm in the Internet of Vehicles from the perspective of data preference in order to improve the data distribution throughput and data transmission coverage and reduce the data transmission delay. The second case is based on data distribution in the IoV with infrastructure. Hot topics of research in these algorithms focus on communication security [15], service latency [16], and other issues in vehicular networking. Literature [17] pointed out that when communicating between vehicles in the Internet of Vehicles, the vehicle can either establish a communication connection with the RSU for data forwarding or rely on nearby vehicles to obtain data. The benefit of this is that the number of vehicles that get their requested data can be greatly increased, thereby increasing the throughput of the network. Literature [18] proposes a mobile content distribution scheme. The scheme consists of two key elements, namely, the Roadside Parking Cloud (RPC) composed of roadside parking vehicles and the mobile cloud (MC) composed of moving vehicles. Then through the mutual cooperation between RPC and MC, the content distribution of the target vehicle is realized. In literature [14], aiming at the influence of vehicle mobility on data download, a cooperative data distribution strategy is proposed to improve the data download rate in one-dimensional two-way highway scenarios. Literature [19] proposes a two-level edge computing architecture and applies it to autonomous driving services. The new scheme prioritizes the transmission of data from the base station to the vehicle and then realizes data distribution through V2V. In order to optimize the network load problem in the Internet of Vehicles, literature [20] relieves the burden of the cellular network by utilizing RSU and vehicle forwarding. The RSU first prefetches appropriate data from the remote network and then obtains a distribution decision with an approximate optimal algorithm to transmit the data to the target vehicle.

Analysis of the above research reveals that the implementation of vehicular communication still faces challenges and difficulties in various aspects. Existing typical D2D cooperative data distribution algorithms are usually classified into two types: unicast mode [21] and multicast mode [22]. The former refers to the fact that, in any given D2D data transmission communication, each terminal can only distribute packets to other terminals within a single cluster. In other words, several independent D2D unicast transmissions must be performed in order to ensure that any one terminal within a cluster can complete the task of forwarding packets to all other terminals within the cooperative cluster. The latter means that the source terminal is able to forward data to multiple target terminals at once. Both the unicast and multicast forwarding algorithms enable information sharing within a cooperative cluster for the purpose of data distribution. However, they do not take into account the variability of D2D link capacity, and the data forwarding rate is easily limited by the few poor D2D links, thus not maximizing the use of valuable spectrum resources. Therefore, aiming at the problem of data cooperative distribution in the context of D2D IoV, this paper proposes an efficient D2D cooperative forwarding algorithm based on multihop relay, including multicast and unicast modes. The new algorithm replaces low-rate one-hop transmission with high-rate multihop transmission by selecting appropriate terminals within the D2D cooperative cluster as relay nodes for packet forwarding and appropriately increasing the number of packet forwarding hops. The innovation of the work is that, in the process of each source terminal forwarding data to all other terminals in the cluster, the algorithm can adaptively search for the optimal relay node, forwarding route, multiple broadcast object, and transmission hop count. Furthermore, by making full use of the multichannel diversity gain in the cooperative cluster, the purpose of improving the forwarding rate and increasing the data distribution throughput is achieved. Simulation experiments prove that the scheme has high global search capability and efficiency, and the new algorithm has significant advantages in improving transmission efficiency and reducing transmission delay compared with the traditional one-hop unicast and one-hop multicast algorithms.

2.1. Introduction to D2D Communication Technology

As one of the key technologies for 5G communication, D2D is a communication technology that allows communication terminals to establish communication connections directly with each other without the need to go through a base station or other core network forwarding to achieve data interaction [23]. The reason for the low efficiency of traditional cellular communication systems is that all communication is carried out without the involvement of a base station. Even when two terminals in close physical proximity communicate, data must be forwarded through the base station, which greatly reduces data transmission efficiency and network throughput. D2D technology is an excellent solution to this problem. In a D2D communication system, every two D2D terminals can be paired into a D2D communication pair, and the two terminals in each communication pair can establish a direct communication link between each other without the need to go through a base station for data forwarding. Therefore, the advantages of the D2D communication technology are as follows: (1) It is helpful to improve the transmission rate of terminal communication and effectively alleviate the problem of data delay. (2) It can help reduce the load of the base station and relieve the pressure on the core network. (3) It is helpful to improve spectral efficiency and system throughput. (4) It can help meet individualized business needs and improve service quality. (5) It has the function of multihop relay, which helps to expand the coverage of the communication system. (6) Compared with short-range communication technologies such as Bluetooth and Wi-Fi Direct, D2D technology is more flexible and can support both licensed and unlicensed spectrum. Figure 1 illustrates the network architecture of D2D communication.

As can be seen in Figure 1, terminal A and terminal B can access the cellular network for data transmission in three ways. The first way is that terminal A and terminal B perform unified scheduling and control by means of the cellular network EPC so as to realize the communication connection between each other. The second way is to establish a direct communication connection between terminal A and terminal B without the involvement of the EPC. This eliminates the need for the cellular network to forward data between terminal A and terminal B, thus increasing the throughput of the D2D network. The third way is to use the relay function from the terminal to the network, that is, the multihop function. The terminal is used as a relay node and is connected to the cellular network via multiple hops, thus enabling the data to be exchanged between the two.

2.2. Internet of Vehicles

With the help of 5G technology, Internet of Vehicles has become one of the representative applications of D2D technology in the field of IoT. The IoV refers to the realization of all-round network connections between vehicles and people, vehicles and vehicles, vehicles and infrastructure, and vehicles and application servers in accordance with the prescribed wireless communication protocols and data transmission standards with the help of a new generation of communication technology. Through the data interaction between them, it provides real-time and efficient network communication support for vehicles and users, thus improving traffic efficiency and providing users with intelligent, comfortable, safe, and energy-saving integrated services. In short, IoV is a mobile communication system that interconnects vehicles with everything else. In view of the fast driving speed of vehicles, frequent network topology changes, and complex and changing traffic environment, it needs to meet various technical requirements such as direct communication between vehicles and nearby terminals, high mobility, and low transmission delay. For the technical requirements of IoV, there are several major application types: V2V (Vehicle to Vehicle), V2P (Vehicle to People), and V2I (Vehicle to Infrastructure). The basic architecture of IoV is shown in Figure 2.

As shown in Figure 2, vehicles equipped with an On-Board Unit (OBU) are connected to everything (including people, vehicles, and infrastructure) through D2D technology. They communicate directly with each other to provide safe, reliable, and real-time information services to vehicle users, minimizing the probability of traffic congestion and damage caused by traffic accidents.

3. System Model and Algorithm Description

3.1. System Model

The system model built in this paper is for D2D cooperative data distribution among users in a cluster serving the same LTE base station. The system model is shown in Figure 3. It includes a 5G base station, the requesting data vehicle S, and M collaborative vehicles. It is assumed that each vehicle can have both traditional cellular communication with the base station and D2D communication with other vehicles. Here, the spectrum resources of the vehicle for D2D communication are orthogonal to those of the cellular user, thus reducing cochannel interference with the cellular user. When S sends a data request to the base station, due to the mobility of S and the limited coverage of the base station, link interruption may occur during data transmission. Therefore, it is necessary to select other vehicles that are consistent with the S motion to assist in completing the data transmission. The base station sends different parts of the data to S and the cooperative vehicle, respectively, and the cooperative vehicle sends it to S through D2D communication.

3.2. Cooperative D2D Distribution Algorithm Based on Multihop Relay

In order to solve the impact of poor communication quality links in D2D cooperative clusters on data distribution, we propose a D2D cooperative distribution algorithm based on multihop relay, which can effectively improve the transmission rate of D2D forwarding and reduce the data transmission delay. Figure 4 shows how the one-hop multicast data distribution algorithm and the multihop multicast data distribution algorithm work within the D2D protocol cluster, respectively. Figure 5 shows the working principle of the one-hop unicast data distribution algorithm and the multihop unicast data distribution algorithm within a D2D protocol cluster, respectively. In the one-hop multicast data distribution scheme of Figure 4(a), terminal A needs to complete the task of directly transmitting data to all other terminals in the protocol cluster. Then the multicast rate of this scheme is determined by the link (A, C) with the worst communication quality in the protocol cluster. Therefore, the resource cost to accomplish this task is 5. However, in the multihop multicast data distribution scheme of Figure 4(b), if terminal D is selected as a relay node for two-hop D2D multicast transmission, then the resource cost to complete this task can be reduced to 3.5. In the same way, in the one-hop unicast data distribution scheme of Figure 5(a), if terminal A establishes a communication link directly with terminal C for D2D unicast forwarding, the resource cost is 5. However, in the multihop unicast data distribution scheme of Figure 5(b), if terminal E is selected as the relay to perform two-hop D2D unicast transmission, the resource cost can be reduced to 2. This shows that the D2D forwarding scheme based on multihop relay can fully utilize the multichannel diversity gain within the cluster to improve the transmission efficiency. However, the key to obtaining the multichannel diversity gain is to select the right relay terminal, forwarding route, and number of hops, which together determine the actual efficiency of D2D forwarding.

In addition, suppose that the D2D cooperative cluster is denoted as F = (C, L), where C denotes the set of terminal serial numbers within a cluster, which is a natural number that uniquely identifies the terminals within a cluster, and L represents the set of D2D links within a cluster. For each terminal within a cluster, that is, i, j ∈ C, the symbol (i, j) denotes the D2D link connecting terminal i and terminal j. d(i, j) denotes the distance between any two terminals i and j within a cluster, which can usually be defined as the number of modulation symbols required to transmit 1 bit of information over the link (i, j). For example, if the link between terminal 1 and terminal 2 supports QPSK modulation and 1/3 code rate transmission, then d(1, 2) = 3/2.

On the basis of the above research, the following describes the D2D multicast and unicast cooperative distribution algorithm based on multihop relay.

3.2.1. D2D Multicast Cooperative Distribution Algorithm Based on Multihop Relay

Definition 1. In the D2D cooperative cluster CM = {1,2, …, M}, when terminal i needs to distribute data packets to all other terminals of CM, the set of source terminals can be defined as CS = {i}. The set of receiving terminals is the set of all other terminals in the CM except CS; that is, CR = CM\CS. For any nonempty subset V in CM and any element i in CM, the following definitions can be made, as shown in formula (1):On the basis of the definition in formula (1), the initial threshold value D(0) is calculated as in formula (2).In turn, according to the threshold value D(n) at the nth iteration, the “candidate relay set” and the “two-hop terminal set” can be calculated using formulas (3) and (4), respectively.Then, the optimal relay node can be found in , and the solution procedure is as formula (5).Meanwhile, the resource cost of the two-hop D2D multicast in the above-mentioned search process can be calculated according to formulas (6) and (7), respectively.Based on the results of the above solution, the total resource cost for two-hop D2D multicast under the current path is shown in formula (8).For the threshold value D(n + 1) of the (n + 1)th iteration, its calculation formula is shown in formula (9).Then, we can get the two-hop D2D multicast scheme that consumes the least resources among all the recorded ones . The solution process is shown in formula (10).Finally, and in the one-hop D2D multicast scenario are compared, and the scheme with less resource consumption is selected as the final D2D multicast distribution scheme. The calculation process is shown in formulas (11) and (12).In summary, the detailed steps of the D2D multicast cooperative distribution algorithm based on multihop relay are shown in Table 1.

3.2.2. D2D Unicast Cooperative Distribution Algorithm Based on Multihop Relay

In the unicast cooperative distribution model, all elements in the set of receiving terminals CR are first sorted according to certain rules, assuming that the sorted set is defined as S = {s1,s2 ,… ,sM-1 }. The sorted rule for the elements is defined as follows: for any element sm and sn in S, if m < n, then d(i,sm) ≤ d(i,sn). Next, an empty path tree is generated, terminal i is set as the root node of the tree, and a set CSR = {i} is defined.

Then, the optimal parent node xopt of any element st can be found in the set CSR, and the resource cost of the element st is shown in formulas (13) and (14), respectively.

Repeat the above process, and the minimum resource consumption from CS to CR can be finally obtained, referring to formula (15).

In summary, the detailed steps of the D2D unicast cooperative distribution algorithm based on multihop relay are shown in Table 2.

4. Simulation Testing and Analysis

The focus of the experiment in this section is to set up a simulation scenario for the data distribution optimization algorithm proposed in Section 3 and combine the existing one-hop unicast and one-hop multicast D2D forwarding schemes and the D2D forwarding scheme based on multihop relay. The data distribution performance is evaluated. Meanwhile, the simulation results are used to analyze and compare the effectiveness and correctness of the algorithm proposed in this paper.

4.1. Parameter Settings

This paper uses the MATLAB language for simulation experiments. Without loss of generality, it is assumed that all mobile terminals are uniformly distributed within the D2D cooperative cluster, that the small-scale fading of the D2D link is a slow-varying flat Rayleigh fading model, and that adaptive modulation coding (AMC) is used for the D2D communication between terminals to make full use of the channel capacity. Table 3 shows some of the parameter settings.

Furthermore, with reference to the link adaptive transmission scheme for LTE systems [24], the coded modulation format is divided into a total of 15 levels. Therefore, the valid values of the corresponding intercluster node distance d in the simulation are shown in Table 4.

4.2. Simulation Results and Analysis

The simulation experiments are divided into two parts: comparison of unicast forwarding schemes (including one-hop unicast forwarding and multihop unicast forwarding) and comparison of multicast forwarding schemes (one-hop multicast forwarding and multihop multicast forwarding). Then, use data reception rate and data reception delay as evaluation metrics, respectively. Figures 6 and 7 compare the performance of the multicast forwarding scenario with the two different transmission methods. Figures 8 and 9 compare the performance of the unicast forwarding scenario with the two different transmission methods. The following simulation results are averaged over 1000 runs in a random scenario.

Figure 6 compares the reception rate of the proposed multihop multicast algorithm with the traditional one-hop multicast algorithm in relation to the number of vehicles in the cluster when the vehicles in the cluster request a stream. The results show that the data reception rate of the multihop multicast algorithm is much better than that of the traditional one-hop multicast algorithm in the simulated D2D cooperative cluster. Furthermore, as the number of vehicles in the D2D cooperative cluster increases, the advantage of the multihop multicast data distribution algorithm becomes more apparent. Figure 7 shows the data reception delay versus the number of vehicles in the cluster for the multihop multicast algorithm and the one-hop multicast algorithm. Compared with the traditional one-hop multicast algorithm, the multihop multicast algorithm comprehensively considers the gain and interference brought by the relay vehicle. Selecting the optimal relay vehicle for data distribution can increase the transmission rate while ensuring that the interference is within the acceptable range. Therefore, the transmission delay is minimized on the premise of ensuring the quality of service.

The reason is that although the traditional one-hop multicast algorithm can make the best use of the “multicast gain,” that is, the data forwarding to all receiving terminals is completed only through one multicast, the transmission rate of multicast is severely limited by the worst D2D transceiver link. Therefore, the transmission rate and transmission delay are easily affected. The multihop multicast algorithm is different, and it selects the optimal relay node, the optimal multicast receiving object, and the optimal transmission hop number adaptively. In this way, while obtaining the “multicast gain,” it also makes full use of the multichannel “diversity gain” in the D2D cooperative cluster. Therefore, the advantages of this algorithm are also more obvious.

Figures 8 and 9 compare the performance of the multihop unicast and one-hop unicast algorithms. The results are similar to the simulation of the multicast algorithm. In all simulated D2D cooperative clusters, the multihop unicast algorithm is significantly better than the traditional one-hop unicast algorithm in terms of data reception rate and data transmission delay. The reason is that the multihop unicast algorithm does not need to repeatedly modify the resource consumption function of each node in each cycle by sorting the elements in the receiving terminal set CR, thus greatly improving the data receiving rate. On the other hand, the algorithm limits the number of D2D forwarding hops within two hops, thereby effectively reducing the delay and signaling overhead of D2D forwarding and ensuring full utilization of the multichannel “diversity gain” in the cluster.

5. Conclusion

As the concept of IoV becomes more and more popular, the existing communication technologies are unable to meet the needs of future IoV for high speed, low latency, high reliability, and mass streaming media distribution. In this paper, we combine the advantages of D2D communication technology and apply it to IoV, which can effectively alleviate the above challenges and bring possibilities for the popularization of IoV services. The key to improving the efficiency of data distribution in cellular networks using D2D communication lies in the design of efficient interterminal forwarding algorithms. In traditional forwarding schemes based on one-hop multicast or one-hop unicast, the forwarding rate is severely limited by the number of D2D links with poor channel quality. To address this problem, this paper proposes an efficient D2D cooperative forwarding algorithm based on multihop relay, including both multicast and unicast modes. By adaptively selecting the optimal relay, route, receiving object, and transmission hops, the algorithm can fully exploit the multichannel diversity gain within the D2D cooperative cluster. Simulation results show that the proposed scheme is far superior to conventional forwarding schemes. The new algorithm shows superior performance in terms of data reception efficiency and data reception delay and has a strong global search capability. The proposed multihop-based relay strategy significantly improves data reception rate and network throughput. The advantages of this algorithm become more apparent as the D2D cooperative cluster grows. Currently, the D2D communication-based data distribution technology for IoV is being improved, but there are still many issues to be solved. As D2D communication needs to be carried out under the control of the cellular system, excessive intracluster data forwarding may increase the time delay of the data distribution service and the signaling overhead of the cellular system. Therefore, the above algorithms limit the number of packet forwarding hops while improving the D2D transmission efficiency. That is, from any source terminal to its target receiving terminal, the number of hops for D2D transmission (unicast or multicast) does not exceed two hops. Therefore, how to reasonably select cooperative vehicles and quickly optimize the cooperative scheme to achieve the goal of maximum energy efficiency of the requested vehicles is the next direction of discussion in this paper.

Data Availability

The labeled data set used to support the findings of this study is available from the corresponding author upon request.

Conflicts of Interest

The authors declare that there are no conflicts of interest.


This work was supported by the Department of Electronic Information Engineering, Leshan Vocational and Technical College, and Huawei Technology Co., Ltd.