1. B5G-6G Future Communication Technology Research Institute

The application technology of wireless transmission around uh-frequency band large-scale MIMO system on UL/DL-Uu and Sidelink-PC5 air interfaces, combined with 5G-6G ecological application scenarios, is carried out to conduct research on the key communication technologies of 3GPP-R18 to R20 that support future smart connected network systems. Including multi-hop broadcast and multicast technology in massive MIMO, hybrid precoding technology, mmWave-terahertz's massive MIMO RF system and channel state synthesis technology, millimeter wave-terahertz's large-scale MIMO wireless positioning technology, RAN and CT collaborative real-time fine-grained resource scheduling technology based on intelligent slice QoS control, intelligent interference avoidance technology in the integrated network of heaven and earth, L2 cognitive multi-hop in 6G new air outlets in Uu and Sidelink Unlimited expansion technology for energy efficiency and spectral efficiency.

5G+AI Upgrade:

Wireless access: 2-12Gbps

Latency benchmark: <10-30ns

Access quantity: > tens of millions

Topology dynamic multi-hop intelligent reconstruction; Multi-source collaborative panoramic learning; Self-awareness and self-management; Real-time distribution and collaboration of endpoint and edge computing.

People (smart phones, wearables), cars (smart cars, rail transit), things (sensors, smart terminals), service robots, intelligent connections of everything

Next Generation X2X Internet (Consumer), Future Smart and Healthy City (Government), Smart Factory and Farm (Enterprise)

2. Institute of Middleware and Operating System of NSAI Service

Based on the Networking System of AI, it creates a self-evolving operating mechanism, core algorithm library and application development ecology, combined with efficient real-time blockchain technology, to form relevant intelligent service platform interfaces and data interaction standards, protocols, and security specifications, etc., to achieve technological innovation from "artificial intelligence" to "self-evolving artificial intelligence". The main research content includes open source software for autonomous controllable chips - including operating systems, operating system middleware, artificial intelligence self-evolving protocols and software, data exchange standards and platforms, etc.

Common problem: Break through the bottleneck of the card neck and create an operating system of Networking System of AI with independent intellectual property rights.

Expected results: Based on distributed artificial intelligence algorithms, intelligent network chips and operating systems, a new kernel based on intelligent streaming processes is provided.

3. Networking System of AI Chip Research Institute

Research independent controllable chips, chip heterogeneous integrated design, packaging and testing, promote the research and industrialization of intelligent terminal system-on-chip, and provide a key technology implementation basis for the deep application of the Internet of Things in smart city construction. The research is carried out from the three levels of system-on-chip, data fusion, and terminal verification, which corresponds to the major scientific problems in lightweight computing architecture, multi-source information fusion, and multi-sensor application scenario analysis. The focus will be on the research of lightweight cloud-side-end collaborative processing computing architecture technology, high-density chip packaging and system-on-chip integration technology, synaesthesia memory-computing integrated design technology, multi-heterogeneous intelligent flow topology design technology and customized and efficient instruction compilation method, so as to support the development of independent and controllable lightweight intelligent terminal processing platform system.

Common problem: build chips, build core chips embedded in intelligent terminals, intelligent networks and intelligent platforms.

Technical breakthrough: Heterogeneous integrated chip design; integrated design of communication, sensing, storage and computing; multi-heterogeneous intelligent flow topology structure; customized and efficient instruction compilation method.

Expected results: Networking System of AI chip

Based on the principle of Networking System of AI; dedicated acceleration chips and devices for Networking System of AI.

4. Ubiquitous Brain Networking Institute

Focusing on distributed artificial intelligence methods and new-generation mobile communication networks, research on brain-computer interfaces and health-assisted rehabilitation training systems. The main research contents include the influence of communication network bandwidth and delay on information processing and learning in distributed artificial intelligence systems，optimization of application service quality through distributed artificial intelligence system scheduling and control of large-scale wireless network resources, self-learning evolution and replication of large-scale distributed artificial intelligence networks and systems, and research on brain neural radio wave control robotic arms.

     1). Intelligent service model based on sub-scenario, IoE intelligent networking terminal that independently runs distributed artificial intelligence algorithm, and trains multi-source information collection, processing and real-time artificial intelligence algorithm;

    2). Service resource scheduling by intelligent service system, < Edge server-Intelligent services> resource scheduling and division of labor;

    3). High real-time perception and processing of multiple information, real-time iterative learning of "brain networking";

    4). Brain network research; Brainwave Behavior Analysis and Deep Neural AI Network + UBN Applications: Autonomous Driving, Safe Counter-Terrorism, Crowd Events, Precision Advertising - > Group Brain Fusion.