Smart technology trends
Deep learning tends to be transparent, explainable, and causal.
Perception Level : Speech Recognition, Face Recognition and Object Recognition.
Cognitive Level : Natrue Language, Machine Translation, Video Semantics and Knowledge Understanding.
Challenges : Privacy Security, Data Protection & Soverelgnty, Ethics and Governance and Jobs.
The third of traditional computing and communication paradigms. Two principles : Shannon's principle, von Neumann architecture, and Moore's law.
New computing system and communication architecture :
- 新sensors：camera, Lidar, ultrasound, radar, microphone array, optical…
- 新dataflow model, 新计算和算法：high bandwidth hierarchical memory, Massive data parallelism, Differential and gradient optimization, Linear algebra accelerator and Boolean function.
- New architecture and chips: CPU/GPU/FPGA/ASIC
- New architecture: 5G and edge intelligence
5G cloud platform, intelligent Internet of things, and edge computing
The main improvements of 5G:
- Space (base stations): imporved from single data stream to hundreds of MIMO antennas
- Time (connection latency): reduced from 170ms to 10ms or even 1ms possible
- Frequency (digital control) increases from MHz in 3G to GHz in 4G and even THz in 5G
- Overall, data rate increases from 2 Mbps to 100 Mbps to 20 Gbps or even 100 Gbps possible from 3G to 5g
5G should provide a virtual network , the left is connected to IoT, and the right is connected to applications. This makes it unnecessary to connect IoT and applications one-to-one.
Then Professor Huang Kai introduced the cloud platform of CHUK-SZ. The bottom layer is IoT sensors, the top is edge servers, and the top is connected to the cloud through the management platform. There are storage pools, big data pools and AI computing pools.
The features and resources are introduced in detail in the middle.
Some scenarios of edge AI computing and communication:
- Self-driving cars: low latency, high mobility;
- ICU patient monitoring: delay sensitive;
- Surveillance camera: anonymous data;
- Smart home: heterogeneous devices and data perception.
Diversity calculation
The explosion of multiple business scenarios and data has promoted more and more diversified computing power supply.
- A variety of business scenarios, differentiated business requirements, require a variety of computing power options. For example, cloud games require CPU to run games and GPU for image rendering; weather forecast scenarios: cloud image analysis requires CPU, and numerical calculation requires GPU.
- The rapid growth of unstructured data drives the development of diverse computing.
High concurrency and low-latency business requirements are driving applications to accelerate to distributed.
Three collaborative innovations:
- Collaborative innovation of computing, storage and network.
- Collaborative innovation of general computing and AI computing.
- Collaborative innovation of software and hardware.
Cloud Edge Collaboration:
| Edge Devices | Edge Computing | Cloud & Data Center | |
|---|---|---|---|
| AI | Perform operation | Reasoning and abnormal data collection | Large-scale learning generation algorithm |
| Big Data | Perform operation | Gateway preprocessing and loading | Big data overall analysis |
| AI | Perform operation | IOT individual analysis and data filtering | IOT overall analysis |
5G edge computing application scenarios: smart manufacturing, smart ports, smart security, smart transportation, smart healthcare, smart education, cloud gaming, cloud video.
At present, everyone is still paying more attention to edge intelligence: that is, how to perform some intelligent AI calculations at the edge, and there are many scenarios. Federal learning is also a hot topic.