Time-Slotted Optical Network Reducing ASIC Switching Bottlenecks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high data rate transmission in data center networks is hindered by the limitations of ASIC switches, including the difficulty in downsizing CMOS transistors, high power consumption, and the need for OEO conversion, which can be costly and inefficient.
Innovation Solution
Implementing an optical network with a full mesh configuration and arrayed waveguide gratings, where nodes are grouped and only addressed during specific time slots, reducing the need for electrical switching and OEO conversion, and using optical switching in the core while maintaining electrical switching on the periphery.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If ASIC switches are used to increase switching capacity, then data transmission capability is improved, but transistor downsizing becomes increasingly difficult and costly
Solution Approach 1:
The patent replaces electrical switching with optical switching in the core network portion. Optical switches use light-based mechanisms instead of electrical transistors, eliminating the need for continuous transistor downsizing while maintaining and potentially increasing switching capacity. This substitution addresses the manufacturing difficulty by moving from semiconductor physics limits to optical physics principles.
Solution Approach 2:
The patent divides the network into two distinct portions: an optical core portion for high-capacity switching and electrical access portions for connectivity. This segmentation allows each portion to use the most appropriate technology for its function, with the optical core handling the bulk of high-speed traffic without being constrained by electrical switching limitations.
2Speed
If pluggable transceivers are used for high data rate transmission, then data rate capability is improved, but data density is limited
Solution Approach 1:
The patent co-packages transceivers with ASIC switches in the access portion, merging previously separate components into an integrated unit. This integration increases data density by reducing the physical space and interconnection requirements between transceivers and switching elements, while maintaining high data rate capabilities through optimized optical-electrical interfaces.
3Quantity of substance
If co-packaging of transceiver and ASIC switch is implemented, then data density is improved, but power consumption remains a problem
Solution Approach 1:
The patent replaces electrical switching with optical switching in the core portion, eliminating the need for high-power electrical signal processing for core network traffic. Optical switching consumes significantly less power than electrical switching at equivalent data rates, thereby reducing overall network power consumption while maintaining high data density through co-packaging in the access portion.
4Use of energy by moving object
If optical switching is introduced to avoid OEO conversion, then power consumption is reduced, but network structure complexity increases
Solution Approach 1:
The patent implements a hybrid network structure with an optical core portion and electrical access portions, clearly segmenting functions to balance complexity and performance. The optical core handles high-speed traffic with minimal OEO conversion, while the electrical access portions provide familiar interfaces and protocols. This segmentation reduces overall complexity compared to a fully optical network while achieving significant power savings.
Solution Approach 2:
The co-packaged transceiver-ASIC units in the access portion serve as intermediaries between the optical core and electrical network devices. These intermediaries perform necessary OEO conversion only at the network edges, minimizing the number of conversion points while maintaining compatibility with existing electrical networking infrastructure.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach simplifies node configuration, reduces power consumption, and scales with the network while maintaining high data transmission rates, addressing the limitations of traditional ASIC switches.
Implementation Method 1
each of the plurality of nodes includes "m" arrayed waveguide gratings (AWGs) in which a plurality of input ports of the AWGs receive corresponding optical signals from the at most "m" nodes belonging to a same group, and a wavelength to be used by one or more source nodes of the at most "m" nodes is set to be compatible with an operating wavelength of the plurality of input ports
Data Source
AI summary
There has been a problem with nodes that are present on the periphery of a flat optical network in that an enormous number of receivers are needed, which have reached the limit of the capacity of the ASIC switch. An optical network of the present disclosure proposes introduction of a slight time-domain limit to data transmission from network nodes to the same destination node. The network operates in accordance with a time slot system for data transmission/reception. The ASIC switch has a switching capacity corresponding to an average volume of incoming traffic at a plurality of nodes, and is provided with a storage medium that stores therein and handle a volume of traffic exceeding this switching capacity. A decreased transmission bandwidth between nodes due to the time-domain limit can also be improved by using a plurality of time slots, and transmitting an optical signal according to optical circuit switching by using an unassigned time slot.