Mobile Edge Intelligence and Intelligent Vehicle Internet Computing

HONG KONG, May 10, 2022 (GLOBE NEWSWIRE) — WIMI Hologram Academy, in partnership with the Holographic Science Innovation Center, has released a new technical paper detailing their exploration of edge intelligence and edge computing in the vehicle internet field. This article follows below:

In today’s world, the application of virtual reality in the field of car internet has become more and more important. Compared to traditional networks, the design of vehicle internet caching algorithms is more challenging, mainly due to the high mobility of vehicles, the frequently changing content requirements and the harsh communication environment.

Edge caching technology is used on the vehicular internet to help deliver content by storing or demanding content on the edge server. Scientists from the WIMI Hologram Academy of WIMI Hologram Cloud Inc. (NASDAQ: WIMI), discussed the application of edge intelligence and edge computing in the automotive internet field, which spans two different scenarios depending on the role of the vehicle, including research on vehicles with different roles (i.e. VaaC and VaaS) .

In the VaaC scenario, the vehicle acts as a cached connected vehicle content consumer to access the required content from the edge server. In the VaaS scenario, the vehicle also acts as a content provider, caching the content in its storage unit. We focused on caching content, namely deciding what content to cache, and divided existing studies into three categories: 1) The perceived time location cache, taking into account the temporal change of importance/ popularity of the content; 2) the spatial location perception cache, which takes into account the different importance/popularity of the same content in different regions; 3) the mobility perception cache, which reduces the impact of vehicle mobility on content cache and delivery.

1. VaaC

As the edge servers are widely deployed, timely content delivery services are available to vehicles passing through their coverage areas. Representative studies from traditional vehicle internet to vehicle smart internet are discussed in detail below.

1.1Cache of the observed time position

In the edge caching system, temporal locality involves two aspects: the freshness of the cached content and the time change of user requests. Given the characteristics of the communication between the RSU (RoadSide Units) and the vehicle, the temporal data transmission problem is described as an NP hard problem. The authors then develop a heuristic scheduling algorithm based on user request requirements (e.g., time limit) to improve the request service’s capability.

1.2 Spatial position-aware cache

In the vehicle environment, certain information, such as traffic information, is related to the location of the vehicle. Therefore, the content freshness of the Internet of vehicles on different roads may vary. For example, there is a deep learning-based caching solution to optimize the caching decisions for the Smart Internet of Vehicles, designed to reduce entertainment content delivery delays. In this scheme, the age and gender of passengers are detected by CNN (Convolutional Neural Network) and multilayer perception (MLP) is used to predict the correct content, storing the content on a specific regional edge server. The vehicle then determines, based on the k-means algorithm and binary classification, which content is accessible from the edge server.

1.3 Mobility perception cache

Delivering a large capacity of content (such as video, music and HD maps) to moving vehicles is challenging due to limited network capacity and intermittent connectivity. In order to minimize vehicle download time, the problem of posting large content to edge servers was studied. This article develops three algorithms to reduce the effect of vehicle mobility on cache performance. Others have proposed a caching strategy to minimize the latency of caching services across multiple Edge Information (EIS) systems. In particular, vehicle mobility is predicted by a long, short-term memory (LSTM, Long Short-Term Memory) complex on a time series. Based on this work, an active caching strategy is developed using a deep reinforcement learning algorithm. Integrating edge caching and computing in EIS is a new research direction to address the issue of vehicle mobility and longevity. In this case, an important issue is how to effectively allocate the limited resources. For example, for the problem of resource allocation in the integrated architecture, two joint optimization models are prepared to determine the optimal cache and computational decisions, and then solved on the basis of deep learning.

2. VaaS

Edge server-based caching is limited by coverage and unreliable connections to the vehicle. Moreover, caching content on a moving vehicle is a good solution. By leveraging vehicle mobility, edge caching can provide more cost-effective and practically enhanced services. Existing studies related to this direction are as follows:

2.1 Cache of the observed time position

For vehicle caching, due to limited onboard storage resources, the temporary location of the content will affect not only the cache services, but also the implementation of other functions on the vehicle. Therefore, determining how long the content will be cached is an important question. Previous methods of content size edge caching have been suggested. We propose a new caching method, Hamlet, to generate content diversity between adjacent nodes by determining the frequency of cache updates for large and small capacity content. Based on this scheme, users can receive different content from adjacent cache nodes in a short time, improving cache efficiency.

2.2 Spatial position-aware cache

Due to the flexible mobility of vehicles and multi-hop data transfer, VaaS mode will improve location-based cache performance. By dividing urban areas into multiple hotspots based on dynamic mobility and vehicle density, historical data is matched locally to predict the future vehicle trajectory. The best utility of caching services can be achieved by integrating vehicles that frequent these hotspots into a cooperative caching system. To mitigate the impact of cellular and communications vulnerabilities on cache services, there is also a dynamic in-vehicle cache relay strategy. The hotspot area contents can be maintained by the cache schedule and vehicle communication.

2.3 Mobility perception cache

The predictable mobility of the vehicle can be leveraged to improve the efficiency of cache-assisted content delivery. Mobile-aware caching in traditional device-to-device networks has been well studied and these methods have recently been extended to vehicle networks. In previous studies, scientists have explored a new caching service where content stored in a vehicle can be queried by mobile or static user requests within communication range. In this scenario, the relationship between caching vehicles and mobile users is key to designing caching strategies.

3. Cached Apps

In addition to the typical content sharing and delivery services, there is great interest in developing new applications supported by edge cache servers. The next one starts with cache-assisted detection and positioning, then introduces other applications in the Internet of vehicles and intelligent transportation systems.

3.1 Cache-assisted observation and localization

Cache assist perception includes automatic overtaking, cooperative collision avoidance, perspective, bird’s eye view and other functions, where the edge cache provides the vehicle with driver assistance and improves the content of the perception of road safety. On the other hand, the cache-assisted location includes Vulnerable Road User Detection (VRU), where the edge cache improves collaboration between RSU, vehicles and pedestrians by caching location information.

3.2 Other applications

Edge caching is constantly emerging on the vehicular internet and some of these uses are described below.

(1) InfoRank: An information-based InfoRank algorithm has been developed for efficient city perception. The proposed algorithm selects and ranks some intelligent vehicles to perform the urban sensing task. Therefore, the monitoring of the environment of these vehicles can be completed at a very low cost. In this algorithm, the vehicle stores the detection data as a data cache server, reducing the load on the edge server.

(2) Over-The-Top (OTT): A new OTT content preview system can be designed by implementing the edge caching mechanism. The vehicle and RSU connections are predicted based on a real test platform. In addition, a content prevalence estimation scheme is proposed to estimate user content requests. After this, the user-requested content is proactively prefetched on the edge server.

(3) Safety information sharing: Data sharing is an effective method that can reduce the data loss caused by the unreliable sensor system and solve the problem of the limited sensing range of autonomous vehicles. Therefore, data security becomes an important task and being able to design a secure information sharing system for autonomous vehicles. The system is designed to improve data security in two scenarios: spreading false data and tracking vehicles.

(4) Traffic management: To analyze the impact of edge caching on traffic control, previous research proposed a traffic control scheme based on edge cache. Traditionally, it is difficult to obtain the optimal state of the traffic system. Therefore, the optimal state of the traffic network contradicts the user equilibrium. In order to reveal the relationship between the user balance and the optimal state of the system, a communication cost model for cache-compatible vehicles is proposed. With this scheme, the traffic networks can be optimized from the communication aspects using the edge cache.

Founded in August 2020, WIMI Hologram Academy is dedicated to holographic AI vision exploration, exploring basic science and innovative technologies powered by human vision. The Holographic Science Innovation Center, in partnership with WIMI Hologram Academy, is committed to exploring the unknown technology of holographic AI vision, attracting, gathering and integrating relevant global resources and superior forces, and advancing comprehensive innovation with scientific and technological innovation as the core, and carry out basic scientific and innovative technological research.


Holographic Science Innovation Center

Email: pr@holo-science. com

Leave a Reply

Your email address will not be published.