Kelly networks are defined as multi-class networks of queues with Poisson arrivals and service times. They capture a broad array of queue service disciplines, including FIFO, LIFO, and processor sharing. One of their most appealing properties is that they give rise to product-form steady-state distributions: the joint distribution of all queues in the network can be written as the product of the marginals over each queue and enables closed-form characterizations e.g., expected queue sizes, packet delays, and server occupancy rates in steady-state. This allows easy assessment of the effects of different routing and scheduling policies on such properties.
Technology Overview

In this invention, Northeastern researchers studied similar issues from the perspective of caching, in particular, Kelly networks in which nodes are equipped with caches, i.e., storage devices of finite capacity, which can be used to store objects.

Exogenous requests for objects are routed towards nodes that store them; upon reaching a node that stores the requested object, a response packet containing the object is routed towards the request source.

As a result, object traffic in the network is determined by both the demand and, crucially, by where objects are cached. This abstract setting can be used to model various networking applications involving the placement and transmission of content. This includes information-centric networks, content delivery networks, web-caches, wireless/femtocell networks, and peer-to-peer networks.

- The method determines caching decisions taking into account congestion via arbitrary convex cost functions, in contrast to prior art that only operates under linear costs
- Reduced computational complexity
- Highly effective in both caching decisions and computational complexity
- Content delivery networks
- Information-centric networks
- Peer-to-peer networks
- Cloud computing 
- License
- Research collaboration
- Partnering
Patent Information:
For Information, Contact:
Myron Kassaraba
Director of Commercialization
Northeastern University
Edmund Yeh
Efstratios (Stratis) Ioannidis
Milad Mahdian
Armin Moharrer