5G is a more advanced mobile communications network deployed in 2018 and later, which mainly includes the following technologies: the millimeter wave technology [1] (26, 28, 38, and 60 GHz) that is able to provide a transmission rate as high as 20 Gbit/s; the massive MIMO technology that can provide “a performance that is 10 times that of the 4G network” for the 5G communications network. As another important technology for 5G, “the low- and medium-frequency band 5G” (5G New Radio) that leverages the frequencies ranging from 600 MHz to 6 GHz, especially 3.5 to 4.2 GHz. Extended and evolved from 4G communications, 5G that represents the development tendency of new generation information communications is going to penetrate every field in the future society; thus, it will construct an omnidirectional user-oriented information ecosystem. This chapter prospects the future application scenarios and hardware development from three aspects: server, mobile terminal, and edge computing, which correspond to the subsequent sections.
7.1 Prospect of Server-Side Applications
7.1.1 Outline of 5G Communications Characteristics
Requirements and characteristics of 5G communications
On the premise of ensuring even improving the quality of service (QoS) for communications, high data rate, low latency and low power consumption are the most essential requirements. In terms of solutions for the establishment of 5G new radio, the key technologies arise such as massive MIMO [2], millimeter wave bands/visible light transmission, filter-bank-based multicarrier (FBMC) modem, dense networking and heterogeneous network, device-to-device (D2D) and in-vehicle network [3] and onboard network, software-defined networking (SDN) [4], cognitive radio networks, and green communications [5].
According to the report of GSMA, until 2025, 5G network will be commercially used in 111 countries and regions throughout the world. Before the 5G technology is laid in a large scale and provided for consumers, two transitions must be accomplished. First, the mobile operators must upgrade their network infrastructures into 5G equipment. Currently, the primary 5G equipment suppliers are Huawei and Zhongxing Telecommunications Equipment (ZTE) from China, Ericsson from Sweden, and Nokia from Finland. Second, the mobile phone manufacturers need to keep up with the pace to embed 5G wireless signal receivers into mobile phones, making full preparation for the 5G network.
At the early stage of the commercialization of 5G, operators will initiate extensive network construction. Revenues of equipment manufacturers from the investment on the 5G network equipment will become the primary source of direct economic output of 5G [6]. According to the White Paper of the Impacts of 5G on Economy and Society, it is estimated that the network equipment and terminal devices will bring the manufacturers a total revenue of approximately RMB 450 billion yuan in 2020, accounting for 94% of the direct economic output. In 2025, the middle stage of the commercialization of 5G, the expenditures from users, other industrial terminal devices and telecom services will grow constantly, which are expected to rise by RMB 1400 billion yuan and RMB 700 billion yuan, respectively, accounting for 64% of the direct economic output. In 2030, the middle and later stage of the commercialization of 5G, internet enterprises and information service industries related to 5G will become the backbone of the direct economic output, which will increase the economic output to RMB 2600 billion yuan, occupying 42% of the direct economic output.
In light of this, we can conclude that in the near future, the commercialization of 5G will result in a great revolution in the basic manufacturing industry and product substitution in equipment manufacturing industry, shining with extremely high commercial value and investment space. Thus, multiple national equipment manufacturers have devoted substantial human and material resources to industries related to 5G.
7.1.2 Outline of the Server-Side Characteristics
Server is a common name for the type of equipment working based on the network environment, which is usually undertaken by various kinds of computers. Unlike a terminal, a server acts as the control and service center of the network, which serves various terminal devices (usually undertaken by various kinds of computation equipment) that are connected to it; it has a high requirement for the computing performance. The three common server architectures include the cluster architecture, the load balancing architecture, and the distributed server architecture. The cluster architecture refers to integrating multiple servers to handle the same service, and it seems that there is only one server from the perspective of client. One advantage of the cluster architecture is that it can use multiple computers to conduct parallel computations to achieve a higher computing speed. The other advantage of the cluster architecture is that it can use multiple computers to backup, which ensures the proper operation of the entire system even if any machine is broken. Established upon the existing network structure, the load balancing architecture can offer a low-cost, effective, and transparent method to extend the bandwidth of network equipment and server, increase the throughput, strengthen the processing capability for network data, and enhance the network flexibility and availability. The distributed resource sharing server is a theoretical computing model server form that studies the geographic information distributed on the network and the database operations affected; it can distribute data and programs to multiple servers. The distributed architecture contributes to the distribution and optimization of tasks in the entire computer system, overcomes the defects in traditional centralized system where strained resources of central hosts and response bottlenecks occur, and addresses issues such as data heterogeneity, data sharing, and computing complexity in the geographic information system (GIS) of network, which is a significant progress in GIS. To ensure the security of important data, the cluster server architecture is mainly used in communications industry. The load balancing that aims at sharing the access loads and avoids temporary network traffic jam, is mainly applied in electronic business websites. The distributed servers are born to achieve the cross-sector high-speed access of multiple single nodes. At present, the distributed server is the first choice for the purpose like content delivery network (CDN).
As an exclusive communications system that a user sends files or accesses remote system or network through remote links, communications server can simultaneously provide communications channels for one or more users as per the software and hardware capabilities. Generally, communications servers are featured with the following functions: the gateway function that provides connections between the user and the host by converting data formats, communications protocols, and cable signals; the service access function that allows remote users to dial-in; the modem function that offers internal users with a group of asynchronous modems for dial-in access to remote systems, information services, or other resources; the bridge and router function that maintains the dedicated or dial-in (intermittent) links to remote local area networks (LAN) and automatically transmits data groupings among LANs; and the e-mail server function that automatically connects other LANs or electronic post offices to collect and transmit e-mails.
Since the performance of a server is crucial to that of the network, 5G and even beyond 5G communications raise the following requirements on the server: strong data processing capability for handling the access of the data in large flow, high stability, and reliability, have a full-functional system with ensured data security, and etc. As is mentioned in Chap. 1, from the perspective of hardware implementation, the main superiority of the ASIC method for implementing a data processing module is that it is able to obtain the optimum overall merit of performance and power consumption, which can satisfy the sharply rising computation capability required by massive MIMO detection chips, and achieve high throughput, high energy efficiency, and low latency. Nowadays, with the rapid development of mobile communications, the deficiency of flexibility prevents it from being further extensively applied. However, in the processing of compute-intensive data, reconfigurable processors cannot only achieve high throughput, low energy consumption, and low latency, but also exhibit unique advantages in flexibility and scalability. Additionally, benefiting from the reconfigurability of hardware, this architecture is possible to execute system update and error correction during the operation of the system, which poses dominant privileges in extending the service life and guaranteeing the release time of products. Thus, the reconfigurable processor becomes a significant and promising research subject in the development of communications in the future.
7.1.3 Server-Side Application
As the latest standard of the global communications, 5G does not confine its significance to a higher speed or improved mobile broadband experience, instead, its mission is especially to connect new sectors and encourage new services, e.g., advocating industrial automation, large-scale IoT, smart home, and autonomous driving, etc. Correspondingly, these sectors and services have higher requirements for the networks, which are higher reliability, lower latency, wider coverage, and higher security. Therefore, a flexible, effective, and scalable network is in urgent demand to meet different requirements from all walks of life.
Main usage scenarios of 5G communications
Continuous wide-area coverage and high traffic capacity hotspot scenarios
Continuous wide-area coverage and high traffic capacity hotspot scenarios are mainly designed to meet the mobile internet business requirements in 2020 and later, which are also primary traditional 4G scenarios. Continuous wide-area coverage is the fundamental coverage method of mobile communications targeting the assurance of users’ mobility and service continuity to offer seamless and high-speed service experience. Its primary challenge comes from the needs to ensure a 100 Mbit/s higher data rate for users anytime and anywhere, which is more obvious in harsh environments such as base station coverage edge and high-speed moving. The scenarios requiring high traffic capacity hotspot are mainly oriented at local hotspot areas to provide users with ultra-high data rate to satisfy the extremely high traffic density demands on the network, which need to be supported by multiple technologies. For instance, super intensive networking can effectively multiplex spectral resources and significantly promote frequency multiplexing efficiency in the unit area; full spectrum access can make the full use of low-frequency and high-frequency spectral resources to achieve a higher data rate.
Scenarios with low power consumption and a large number of connections, and low latency and high reliability
Main usage scenarios and key challenges of performance for 5G
Scenarios | Key challenges |
|---|---|
Continuous wide-area coverage | 100 Mbit/s user experienced data rate |
High traffic capacity hotspot | User experienced data rate: 1 Gbit/s Peak data rate: tens of Gbit/s Traffic density: tens of Tbit/km2 |
Low power consumption and a large number of connections | Connection density: 106/km2 Ultra-low power consumption and cost |
Low latency and high reliability | Air interface latency: 1 ms End-to-end latency: at millisecond level Reliability: close to 100% |
The specific usage scenarios are introduced as follows.
7.1.3.1 IoV
As far as China is concerned, the national car ownership has reached 217 million up to 2017, which is increased by 23.04 million with a growth rate of 11.85% compared with that of 2016. Moreover, the proportion of automobiles in motor vehicles increases constantly from 54.93 to 70.17% in the recent 5 years; automobiles have become the main part of motor vehicles. In terms of distribution, there are 53 cities in China whose car ownership is more than a million, of which 24 cities amount to 2 million and 7 cities possess more than 3 million, Beijing, Chengdu, Chongqing, Shanghai, Suzhou, Shenzhen, and Zhengzhou. In western areas, the motor vehicle ownership reaches 64.34 million with the fastest growth rate. In 2017, motor vehicle ownership in eastern, middle, and western areas were 155.44, 90.06, and 64.36 million, accounting for 50.17, 29.06, and 20.77% of total motor vehicles in China, respectively. Among them, the automobile ownership of western areas in recent five years rises by 19.63 million with a growth rate of 19.33%, which is higher than the 14.61 and 16.65% of eastern and middle areas [7].
Application of 5G in automobile industry
With the continuous and rapid increase of the motor vehicle ownership, the number of drivers also substantially grows synchronously, driving the annual increment of recent five years to 24.67 million. In 2017, the number of national motor vehicle drivers reached 385 million, of which automobile drivers accounted for 342 million. 30.54 million drivers occupying 7.94% of the total drivers had less than 1 year of driving experience. On one hand, the surge of automobile ownership and the improper management of parking lots aggravate the low utilization rate of parking space, which raises the “parking problem” to be solved. On the other hand, the high requirements on motor vehicles and sharp increased inexperienced drivers are endangering the traffic safety. Therefore, the research, development, and upgrade of the driving assistance even self-driving technologies are imminent.
Architecture of the cloud-based parking management system
On the condition of satisfying the performance requirements such as power consumption, latency, and throughput, to cope with the increase of motor vehicle ownership by leaps and bounds, the data processing scale of massive MIMO signal detection chips will be certainly raised. In this scenario, the reconfigurable architecture is more suitable compared with the customized ASIC architecture.
Autopilot diagram of Tesla
In high-speed driving mode, real-time data processing and information interaction are extremely important [9], which is one of the typical usage scenarios of the low delay and high reliability. Thus, the low delay and high throughput are the most pressing performance demands. ASIC does not only have the natural strength in energy efficiency but also have low chip manufacturing cost due to the large volume of motor vehicle ownership (after mass production, one-time engineering cost can be amortized over all chips). Therefore, ASIC-based massive MIMO signal detection chips have an optimistic application prospect.
7.1.3.2 Cloud Computing
Three-layer model of cloud computing
Being the technical and development hotspot of the current Internet, cloud computing integrates infrastructures, application platforms, and application software into a complete network structure [10]. Based on the internet technologies, this system provides external services in self-service and on-demand manners; it is featured with broad network access, virtualized resource pooling, rapid elasticity, measured service, and multi-tenant, posing as an active reference for operators to improve their network application capabilities. According to different service modes, cloud platform can divided into three service modes, infrastructure as a service (IaaS), platform as a service (PaaS), and software as a service (SaaS). Using the technologies such as virtualization and distributed computing, cloud computing incorporates various computer resources into an address pool via a computer network, which is a new type of on-demand service mode [6]. Mobile cloud computing (MCC) has the characteristics including weakened limit for terminal hardware, more convenient data storage, personalized services, and ubiquitous availability [11], which should be supported by large-scale and prompt data quantity processing at the server side. In summary, in this scenario, the ASIC-based massive MIMO has a broad application prospect.
7.2 Prospect of Mobile-Side Application
Mobile computing terminal, which is by definition referring to the computer equipment used during movement, mainly including the wireless onboard terminal and the wireless handheld terminal. Thanks to the rapid progress of broadband wireless access technology and mobile internet technology, people are eager to ubiquitously obtain information and services easily even during movement. As the access interface to wireless network, mobile terminals witness a flourishing tendency with many kinds of mobile equipment (smartphones and pads) springing up. Current mobile computing terminals cannot only accomplish voice chat, voice videos, and photographing but also enable rich functions such as Bluetooth, GPS location, and information processing, which plays an ever more important role in human society. At Mobile World Congress 2018, “5G era” stood out as one of the spotlights. As the fifth generation of mobile communications network, 5G is capable of achieving the “internet of everything.” Compared with 4G communications technology, 5G has a much higher data rate, and achieves significant improvements in stability and power consumption; it will significantly affect the mobile computing terminals. Different from previous generations of communications technologies, the mobile computing terminals of 5G cover a more extensive range, generating many new products such as wearable devices and home networking devices. In addition, mobile computing terminals are more humanized to satisfy users’ requirements at faster information transmission rate. More importantly, 5G lays the foundation for the development of other related technologies because the fast data rate is universally required among big data, cloud computing, AI, and self-driving.
- (1)
Low power consumption. As the most essential part of mobile terminals, baseband chip mainly synthesizes baseband signals to be transmitted and decodes the received signals. During the transmission, it encodes signals into baseband codes that can be transmitted, while it decodes the received baseband codes into audio signals during the receiving. In the smart terminal market, the data processing on the baseband chip of smart terminals is becoming increasingly heavy, therefore, the low power consumption design of the baseband chip is significant to the development of smart terminals [13].
- (2)
Low latency. A growing number of applications raise higher requirements on the path delay. In this case, baseband chip needs to process data in real time with a latency at millisecond level.
- (3)
Low cost. In the fifth generation of ultra-intensive network, the size of a micro base station will be very tiny with short distances between stations. As the deployment density is very high, the cost of micro base stations is very important to the operators. The deployment should cover both the indoor and outdoor scenarios using low-cost CMOS power amplifiers to access nodes ranging from several meters to 100 m.
- (4)
High capacity. Baseband chip needs to accomplish high capacity, energy efficiency, and spectrum efficiency.
As for the requirements of the baseband chip, massive MIMO detection VLSI architecture based on ASIC and the reconfigurable massive MIMO detection VLSI architecture may have a promising application prospect, from which aspects this section will be elaborated.
7.2.1 Application of ASIC-Based Detection Chips
In addition to supporting the mobile broadband development, 5G also supports numerous emerging application scenarios. Increasing numbers of applications promote higher requirements for data transmission, i.e., low latency and high throughput, which demands more for the design of massive MIMO detection chips. ASIC-based massive MIMO detection chips are endowed with the potential to meet the future application requirements in latency and throughput. The applications of ASIC-based massive MIMO detection are illustrated by taking the following future applications as examples.
7.2.1.1 VR and AR
Application of VR in remote education
Subversion of VR in the traditional education
AR-based driving assistance system.
©[2018] IEEE. Reprinted, with permission, from Ref. [14]
The wireless transmission plays a crucial role in VR and AR. For example, the tactile network in VR and AR must process data in real time to meet users’ demands. Under the circumstances where a large number of VR and AR terminals have data to be processed, massive MIMO detection needs to satisfy the requirements of high accuracy, low latency, and high throughput. Therefore, the ASIC-based massive MIMO is prospective particular for cases with the requirements of low latency and high processing rate.
7.2.1.2 Self-driving
Roadmap of driving assistance function.
©[2018] IEEE. Reprinted, with permission, from Ref. [15]
In the massive MIMO detection system, the more optimal VLSI architecture can significantly improve the performance of the detection chip and lower the power consumption and latency of the system, which can better accomplish the real-time communications and achieve a higher safety in the self-driving field. This greatly benefits the reduction of traffic accidents and the improvement of the traffic congestion situation. In the current automobile security scenarios, the reaction time to avoid collisions is shorter than 10 ms, while the bidirectional data exchange of self-driving vehicles may require the latency be within 1 ms, which can be technically realized through tactile network and 1 ms end-to-end latency. Thus, full autopilot technology will definitely change traffic behaviors. In terms of the distance between vehicles, autopilot technology needs to detect the potential safety-critical conditions in advance, which can be supported by the future wireless communications system with high reliability and proactive predication [16]. With the increase of the self-driving terminals, data exchange is required among an increasingly number of users. For the self-driving terminal, how to cope with multiuser requirements and shield multiuser interference is a big challenge. The highly effective massive MIMO detection architecture can meet the high-speed processing requirements in the self-driving system to lower the latency. Meanwhile, massive MIMO detection architecture is capable of transmitting massive data and processing correspondingly to improve the system throughput. For the application scenarios with high interference and noise, the nonlinear massive MIMO detection architecture can enhance the detection precision while maintaining certain latency and throughput, which is very important for the security of self-driving. The ASIC-based massive MIMO detection architecture not only can meet the latency and throughput requirements but also poses certain advantages in power consumption. In addition, considering the high mobility of self-driving terminals, the massive MIMO detection architecture must adapt to distinct scenarios and requests.
The massive MIMO detector receives and restores information, which has great effects in improving the channel capacity and communications efficiency and ensuring instant communications of remote diagnosis. More importantly, a proper massive MIMO architecture can accelerate this process. Thus, designing a more optimal massive MIMO VLSI architecture has always been the research topic of many researchers. In 5G era, most communications systems of mobile terminals are inevitably associated with massive MIMO technology. The massive MIMO can fulfill not only high capacity and speed but also low power consumption and cost, which will contribute to the boost of further flourish of mobile terminals.
7.2.2 Application of Reconfigurable Detection Chips
In the future, more applications will emphasize high energy efficiency as well as flexibility and scalability to adapt to different algorithms, MIMO system scales, and detection performance requirements. To accommodate these features, the reconfigurable MIMO signal detectors have gradually become the hotspot in the academia in recent years. This is because the reconfigurable MIMO signal detectors can fully exploit and utilize the data parallelism in algorithms and dynamically reconfigure chip functions via configuration flow, which can achieve a certain tradeoff between efficiency and flexibility compared with GPP and ASIC. The following sections give examples to show the applications based on the reconfigurable massive MIMO detectors.
7.2.2.1 Intelligent Manufacturing (IM)
IM-related technologies
IM must make the most of communications means at the network layer to control and operate all the intelligent equipment by using wireless communications. In turn, massive MIMO detection in IM equipment must meet the requirements of high stability, flexibility, and scalability. Therefore, how to realize the above requirements will be a challenge for massive MIMO detection chip. Nevertheless, the reconfigurable massive MIMO detector shows certain advantages in these aspects, which is featured with very high potential application values. Moreover, with the popularization of IM, a growing number of industrial intelligent equipment will leverage wireless transmission systems, which raises the issues of upgrade and compatibility for equipment systems. Thus, the precision requirement for the design of massive MIMO detection chips will be increased. Therefore, how to reduce the interference between equipment and the impact of other environmental noise on the signal transmission, and improve flexibility and scalability will be the primary research directions for the design of reconfigurable massive MIMO detectors.
7.2.2.2 Wireless Medical
Wireless medical and monitoring system
In wireless medical, the reliability plays a vital role. To suit distinguished equipment and human characteristics, the hardware circuits with more flexible framework are required. Also, to adapt to the continuously developed and updated equipment requirements, the wireless baseband processing circuits should be scalable to lower the cost. Exactly, the reconfigurable massive MIMO detector will have a bright future regarding to these aspects. In addition, most massive MIMO detection algorithms are with deep parallel computing, and reconfigurable architecture shows its superiority in efficient processing of parallel computation [19]. Generally speaking, the higher the parallelism and the lower data dependency in the algorithm are, the more suitable it is to be accelerated using reconfigurable methods, which is also a reflection of the algorithm at the hardware level. Therefore, reconfigurable massive MIMO detector can effectively fulfill the computation with high parallelism.
7.3 Prospect of Applications of Edge Computing
Typical MEC application scenarios
7.3.1 Concept of Edge Computing
Comparison between MEC and MCC
Comparison items | MEC | MCC |
|---|---|---|
Server hardware | Large-scale data center (each possessing a lot of powerful servers) [10, 25] | |
Server location | Coexist with wireless gateway, Wi-Fi router, and LTE base station [2] | Installed in exclusive buildings with the scale comparable to several football courts [11, 26] |
Deployment | Intensive deployment by telecom operators, MEC suppliers, enterprises, and family users participated and light configuration and plans required [2] | Deployed in a few places all around the world by IT corporations such as Google and Amazon with complicated configuration and plans required [10] |
Distance to end user | Short (dozens of meters to several hundred meters) [26] | Long (probably across continents) [26] |
Backhaul use | Infrequently used, alleviating congestion [12] | Frequently used, causing congestion [12] |
System management | Layered control (centralized/distributed) [13] | Centralized control [13] |
Supported latency | ||
Application | Compute-intensive applications with high requirements on latency, e.g., VR, self-driving, and online interactive games [2, 17] | Compute-intensive applications without high requirements on latency, e.g., online social, mobile commerce/healthcare/study [18, 19] |
- Proximity:
As MEC server is deployed proximal to the information source, edge computing is very suitable for capturing and analyzing the key information of big data. Moreover, edge computing can directly access user equipment; thus, specific business applications are easily derived.
- Low latency:
As MEC server is proximal to or directly operating on the terminal devices, the latency is greatly lowered. This makes the application feedback faster, improves user experience, and dramatically reduces the possibilities of congestion incurred in other parts of the network.
- High bandwidth:
As MEC server is proximal to the information source, it can complete simple data processing without uploading all data or information to the cloud, which reduces the transmission pressure of the core network, decreases network congestion, and enhances network transmission speed.
- Location awareness:
When the network edge is a part of the radio access network, no matter Wi-Fi or honeycomb, local services can identify the specific location of each connection equipment with a relatively little information.
Basic system architecture of MEC
7.3.2 Application of Detection Chips in the Edge Computing
MEC enhancing the integration of data center and 5G
- (1)
Local execution, in which the entire computation process is executed locally at the user equipment without offloading computation to the MEC, e.g., due to that the MEC computation resources are unavailable or the performance cannot be improved by offloading.
- (2)
Full offloading, in which the entire computation is offloaded to be processed at MEC server.
- (3)
Partial offloading, in which part of computation is executed locally while the left is offloaded to the MEC server for processing.
Computation offloading, especially partial offloading, is a very complicated process which will be impacted by multiple factors such as user preference, wireless and backhaul connection quality, user equipment computation capability, or utilizability of cloud computing capability. Application model/category is also one of the significant aspects of computation offloading because it determines whether full or partial offloading fits, which computations can be offloaded, and how these computations can be offloaded [27]. MEC server is able to provide more powerful computation capabilities than user equipment, offloading computation to MEC server for processing can shorten the data processing time and save the energy of terminal devices consumed for data processing. However, we cannot ignore the fact that offloading data to be processed by the user equipment to the MEC server (uplink) needs to consume transmission time and energy so does it when the MEC server transmits the processed data to the user equipment (downlink). When the computation amount of an application is not very huge, especially when the processing capability of user equipment satisfies the requirements, the aforementioned data transmission (uplink and downlink) may waste time and energy, causing the performance loss. Thus, a reasonable mechanism is required to make the decision of whether to perform computation offloading. MEC technology has relatively high requirements on uplink and downlink data transmission, which are mainly reflected in the low latency, high throughput, and low power consumption in massive MIMO detection. ASIC-based massive MIMO detection chips show outstanding performances in these aspects and can be implemented at the MEC terminal, to reduce latency and power consumption, and improve throughput.
MEC terminal management
The MEC server is a small data center, and each data center consumes less energy than a traditional cloud data center. However, its intensive deployment mode causes serious problems in the energy consumption of the whole system. Therefore, it is definitely a key to developing innovative technologies to achieve green MEC. Compared with green communications system, the computation resources of MEC server must be appropriately allocated to realize the required computation performance, making traditional green wireless technologies no longer suitable. In addition, as the past researches on green data communications network has not considered wireless resource management, they are not applicable to green MEC. Besides, the highly unpredictable computation workload pattern in MEC server poses another big challenge for the resource management in MEC systems, calling for advanced estimation and optimization techniques [24]. What’s more, there are increasing demands for secure and privacy-preserving mobile services. While MEC enables new types of services, its unique features also bring new security and privacy issues. First of all, the innate heterogeneity of MEC systems makes the conventional trust and authentication mechanisms inapplicable. Second, the diversity of communication technologies that support MEC and the software nature of the networking management mechanisms bring new security threats. Besides, secure and private computation mechanisms become highly desirable as the edge servers may be an eavesdropper or an attacker. These motivate us to develop effective mechanisms [24]. We can also circumvent some power consumption and security related issues from hardware circuits. The reconfigurable massive MIMO detector is close to ASIC in energy efficiency, and can implement different algorithms and signals processing of different scales, demonstrating high flexibility and scalability. In addition, as the PEs and interconnect muddles inside the reconfigurable massive MIMO detector are relatively regular, it is difficult to obtain the algorithm information by observing the hardware architecture and circuit composition. This feature can improve the hardware security and avoid some MEC security issues.
Next, the practical application of the IoV is used as an example to demonstrate the advantages of MEC. The IoV has special requirements for the data processing. The first requirement is low latency, i.e., to achieve the early warning of collision when vehicles are moving at high-speed, the communications latency should be within several milliseconds. The second requirement is high reliability. For safe driving requirements, the IoV requires higher reliability compared with ordinary communications. Meanwhile, as vehicles are moving at high speed, signals must meet the high reliability requirements on the basis of being able to support high-speed motion. With the increase of networked vehicles, the data quantity of the IoV also grows and as a return, the requirements for latency and reliability are higher. After MEC technology is applied to the IoV, due to the location characteristics of MEC, the IoV data can be saved in places proximal to the vehicles to lower the latency, which is quite suitable for the service types with high latency requirements such as anti-collision and accident warning. Meanwhile, the IoV should ultimately be used to help in driving. The location information of vehicles changes rapidly when vehicles are moving at high-speed. Nevertheless, the MEC server can be placed on the vehicle to accurately sense the location change in real time, which improves the communications reliability. In addition, what the MEC server processes are the real-time IoV data with great values. The MEC server analyzes the data in real time and transmits the analysis results to other networked vehicles in the proximal area with ultra-low latency (usually in milliseconds) to facilitate the decision-making of other vehicles (drivers). This approach is more swift, autonomous, and reliable than other processing methods.