Abstract
In order to improve future network performance, this paper proposes scalable intelligence-enabled networking (SIEN) with eliminating traffic redundancy for audio-visual-tactile Internet in 5G scenarios, such as enhanced mobile broadband, ultra-reliable and low latency communication, and massive machine-type communication. The SIEN consists of an intelligent management plane (ImP), an intelligence-enabled plane (IeP), a control plane, and a user plane. For the ImP, the containers with decision execution are constructed by a novel graph algorithm to organize objects such as network elements and resource partitions. For the IeP, a novel learning system is designed with decision-making using a congruity function for generalization and personalization in the presence of imbalanced, conflicting, and partial data. For the control plane, a scheme of identifier-locator mapping is designed by referring to information-centric networking and software-defined networking. For the user plane, the user-related traffic, which may include identifiers, registrations, requests and named data objects, is forwarded to implement the SIEN and test its performance. The evaluation shows the SIEN outperforms four state-of-the-art techniques for redundant traffic reduction by up to 46.04% based on a mix of assumption, simulation, and proof-of-concept implementation for audio-visual-tactile Internet multimedia service. To confirm the validity, the best case and the worst case for traffic offloading are tested with the data rate, latency, and density. The evaluation only focused on the scalability issue, while the SIEN would be beneficial to improve more issues such as inter-domain security, ultra-low latency, on-demand mobility, multi-homing routing, and cross-layer feature incongruity.
Highlights
Networks and networking exist in order to transfer entities, objects, energy or information as traffic between ingress nodes and egress nodes
EVALUATION IN 5G SCENARIOS the scalable intelligenceenabled networking (SIEN) was evaluated by assumptions, simulation and proof-of-concept (PoC) implementation using information-centric networking (ICN) in 5G scenarios such as massive machine type communications (mMTC), enhanced mobile broadband (eMBB), ultra-reliable and low latency communications (URLLC)
We assume that all globally unique long identifiers (IDs) for objects are with 160 bits fixed lengths as flat names and each ID binds less than five network addresses (NAs) with 32 bits using IPv4 addresses or 128 bits using IPv6 addresses for eMBB, URLLC, and the nonlocal domain of mMTC
Summary
Networks and networking exist in order to transfer entities, objects, energy or information as traffic between ingress nodes and egress nodes. We have to be targeting to satisfy more 5G requirements in terms of scalability for a massive number of objects, high security, low latency, mobility on demand, effective compatibility, efficient manageability, and so on. The identifier could be any object such as mobile devices, data, service or a piece of content, while the locator could be any network addresses. With the increase of terascale (1012), petascale (1015) or even exascale (1018) objects such as data, instead of the existing ten-billion-level (1010) devices, the current networks suffer from the issue to deal with scalability due to the cognitive complexity of automaticity, personalization, and generalization, since it is inefficient for complete data manual processing to handle the large-scale network data in practice.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have