Huawei Fellow Dr. Wen Tong: New Theories and Models Inspired Huawei's Revolutionary Application-driven Network
SHENZHEN, China -- Huawei Fellow Dr. Wen Tong announced that the company's revolutionary Application-driven Network (ADN) was inspired by three information consumption models that support a variety of applications. What makes ADN innovative is its top-level architecture design: Network layers are vertically integrated from the top down, and optimal network resources required for applications are allocated and controlled from end to end.
"The environment in which networks evolve is changing dramatically," said Dr. Wen. "Business models and user needs vary at different stages of network development. The Future Network Theory Lab of Huawei's 2012 Laboratories has discovered that different communications scenarios comply with different information consumption models and lead to different types of network architecture. Huawei prioritizes applications in network evolution. It has developed the ADN concept and completed the analysis of network theory models as well as architecture testing under typical scenarios."
According to Dr. Wen, applications are supported by three information consumption models.
First is the Poisson distribution model, used for voice communications. Under this scenario, most subscribers' call frequencies and durations remain at a mean value. The model is suitable for hierarchical network architecture.
Second is the power-law distribution model, used for Internet communications. Under this scenario, most users are connected to super Internet nodes. This model is a perfect fit for flat network architecture.
Third is the Markov process model, used for machine-to-machine (M2M) communications. In the M2M era, the application needs of industries differ greatly. For example, in the Internet of Vehicles, car-to-car communications require extremely low latency. In telemedicine, remote video systems require ultra-wide bandwidth, low latency, and high reliability. In this case, networks must create small systems locally and huge systems globally. The Markov distribution model supports network architecture with both distributed and centralized controls.
Regarding the innovative ADN architecture, Dr. Wen remarked, "Traditional communications networks adopt a horizontal architecture, with resources scattered at different network layers. They aim to provide coverage by region and maximize the utilization efficiency of resources at different layers. However, these networks are unable to meet end-to-end application needs."
According to Dr. Wen, to address this issue, Huawei's ADN adopts a top-down vertical architecture. Huawei has developed three theories that guide the design of a three-layer ADN: the information consumption theory, information control theory, and information dissemination theory. At the application layer, data modeling is conducted according to the characteristics of different services. At the resource control layer, network resources are coordinated globally and locally to deliver a superior user experience and maximum resource utilization. At the data layer, application decoupling through service layering is achieved to support more efficient resource sharing and coordination."
"ADNs represent a new business theory, which requires us to rethink network operations and technology development from the perspectives of customer value, network value, and carrier value," said Dr. Wen. "ADNs recommend an application- and demand-driven approach to network construction and development, thus reshaping carriers' traditional ways of networking," he explained.
ADNs will open the door to applications, enabling carriers to diversify their traditional traffic-only or bandwidth-only sales models and create new revenue streams. "In the future, carriers will be able to expand their market space by leasing ADNs with enhanced applications to vertical industries, such as transportation, healthcare, education, finance, and the IoT," Dr. Wen continues.
At Globecom 2015 on December 10, Huawei caused a sensation when it first announced its ADN concept.