Low Latency Caching and Forwarding Method for Content Distribution Networks

INV-18030

 

Research on information-centric networking (ICN) architectures over the past few years show a number of central network design issues. One prominent issue is how to jointly design traffic engineering and caching strategies to maximally exploit the bandwidth and storage resources of the network for optimal performance. While traffic engineering and caching have been investigated separately for many years, their joint optimization within an ICN setting was under‑explored.

 

Current networks suffer from latency/slowed response based on the amount of data being transferred and the amount of congestion on a network.

 

In this invention, a new unified framework is used to minimize congestion-dependent network cost in information-centric networks by jointly optimizing forwarding and caching strategies. Since caching variables are integer-constrained, the resulting optimization problem is NP-hard (non-deterministic polynomial-time hardness).

 

These algorithms aim to reduce those delays and provide enhanced network performance. They involve a method for specifying how to forward requests (at routers) for content and place contents in storage devices (caches) in a network. This allows them to minimize congestion-dependent network cost, such as latency. The method simultaneously determines dynamic request forwarding and caching, adaptive to request demand in network. It works for general multi-hop network topologies given any loop-free request routing scheme and can be implemented in a distributed manner.

 

Northeastern researchers have developed a necessary optimality condition and leveraged this result to design MinDelay: an adaptive and distributed joint forwarding and caching algorithm based on the conditional gradient algorithm. 

 

The MinDelay algorithm efficiently yields feasible routing variables and integer caching variables at each iteration and can be implemented with low complexity and overhead. Over a wide range of network topologies, simulation results show that MinDelay typically has significantly better delay performance in the low to moderate request rate regions. MinDelay and VIP algorithms complement each other in delivering superior delay performance across the entire range of request arrival rates.

  

The method is distributed, in that routers and storage devices make decisions on forwarding and caching based on packets passing through them and on communication with immediate neighboring nodes. The method is also adaptive, in that forwarding and caching automatically adapt to changes in content demand.

 

 

- Can minimize congestion-dependent network costs through jointly optimal request forwarding and caching

- It is both distributed and adaptive

- Operates based on packets passing through and communication with immediate neighboring nodes

- At low to intermediate request arrival rates, the method provides latency performance significantly lower than any other known algorithm, in evaluations over a broad array of network topologies