In-Network Compression Using Reference Identifiers for Bandwidth Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing network solutions fail to optimize bandwidth usage by not leveraging intra-packet, inter-packet, and inter-flow redundancies, leading to inefficiencies in wireless access points and increased costs for wired access installations.
Innovation Solution
A method for constructing data structures that associate frequently occurring data segments or strings with reference identifiers, allowing for in-network compression across multiple hops while preserving packet boundaries, enabling transparent bandwidth savings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If end-to-end compression is used, then bandwidth usage is optimized, but inter-flow redundancies cannot be exploited and application transparency is lost
Solution Approach 1:
A network element acts as an intermediary that collects packets from multiple flows, identifies common data segments across flows, and replaces them with references. This mediator enables inter-flow redundancy exploitation while maintaining end-to-end transparency, as the compression occurs in the network rather than at endpoints.
Solution Approach 2:
The invention segments data into identical segments across multiple flows, allowing each flow to be compressed independently while still exploiting redundancies across flows. Packets are divided into segments that can be matched across different application flows, enabling efficient compression without requiring application-layer involvement.
2Loss of energy
If proxy-based solutions are implemented, then some compression is achieved, but per-flow state maintenance and application awareness are required
Solution Approach 1:
The system uses self-service by having the network element automatically collect packets, identify redundancies, and perform compression without requiring external control or state maintenance. The compression process is autonomous, eliminating the need for per-flow state tracking that plagues proxy-based solutions.
Solution Approach 2:
Multiple flows are merged at the network element where compression is performed centrally. This combining approach eliminates the need for separate per-flow state maintenance by handling all flows through a unified compression mechanism that exploits inter-flow redundancies.
3Loss of energy
If in-network compression is implemented, then bandwidth is reduced and wired access installation is minimized, but implementation compatibility with non-INC compliant devices must be maintained
Solution Approach 1:
The invention creates copies of packet data at the network element, allowing compression to be performed on copies while original packets maintain their integrity for compatibility with non-INC compliant devices. This copying approach enables bandwidth reduction without affecting devices that cannot understand INC protocols.
Solution Approach 2:
Compression is applied locally at specific network elements rather than end-to-end, allowing selective compression of certain traffic while maintaining compatibility with devices that cannot handle INC. This local application of compression enables bandwidth reduction while preserving universal device compatibility.
Data Source
AI summary
Media and methods are provided for constructing data structures associating data segments in packets with reference identifiers to optimize bandwidth usage within a network during transmission. In one embodiment, a data structure is initialized to contain no data. A minimum length of a data segment to be examined is determined. For each packet in the transmission, data segments of the packet are examined. Data segments are added to the data structure as necessary. Examination is repeated along with an appropriate addition until the packet contains less data than a segment of the minimum length and further continued until all packets are examined. The data structure is constructed associating the portions of the packet and respective reference identifiers. In certain embodiments, the count can be a total count of frequently occurring segments, the count of unique segments or the count of packets containing unique segments. In other embodiments, the construction is accomplished by a compression library. In still other embodiments, one or more filtering criteria are added to the process to optimize efficiency. Embodiments are described where packet size, packet source, packet destination or proportion of alpha-numeric characters are used as the filtering criteria.