Current commercial cellular networks mostly rely on inflexible hardware-based architectures. While this approach is good enough to provide mobile users with high performance services in many cases, resources such as spectrum, computing and transmission power are often partially or inefficiently utilized due to the inflexibility of the infrastructure
For this reason, over the years the research community has developed a variety of optimized protocols and algorithms to better utilize network resources and improve network performance. Due to the lack of hard real-time constraints, offline solutions to optimize frequency and network planning have been adopted in many commercial network applications. Because of their strict timing requirements, the same does not hold for lower-layer problems, such as scheduling and power control. Policies concerning these rely on static or heuristic solutions. As of today, a fast and flexible programmatic control framework which can jointly meet real-time requirements of lower layers of the protocol stack and autonomously and quickly adapt optimization variables to the time-varying network dynamics (e.g., channel conditions, position of users, size of transmission queues, etc.) has been unavailable.
To address these challenges, research has been focusing on pushing the SDN control plane as close to the edge of wireless networks as possible]. For instance, in CellSDN and OpenRadio, network-level decisions are made by a centralized controller, and latency-sensitive local decisions are executed inside the base stations.
Technology Overview
Software-Defined Networking (SDN) has been proposed as a promising technology to overcome the above limitations by abstracting the physical network and decoupling the data and control planes. There is a need to bridge the gap between Software-Defined Networking (SDN) and cross-layer distributed optimization techniques.
Unlike state-of-the-art  SDN-inspired solutions for cellular networking,  CellOS hides low-level network details through a virtual network abstraction,  allows the Telco Operator to define high-level control objectives without requiring knowledge of optimization techniques or of the underlying network, and automatically generates distributed control programs to be executed at each base station unit to optimize the network performance.
These approaches either neglect optimization or only provide nonscalable centralized solutions. The situation will worsen with the advent of 5G, where Telco Operators (TOs) adopt tiered network architectures with densely-deployed base stations.  TOs need to set network optimization objectives through a few lines of code and without requiring expertise in optimization theory or knowledge of network implementation details.
In this invention, Northeastern Researchers embrace the SDN paradigm to bring the decoupling among control and data planes to its farthest limit by including an additional decoupling layer among control and optimization planes.
CellOS allows Telco Operators to set network optimization objectives through few lines of code and without requiring expertise in optimization theory or knowledge of network implementation details. By using an abstraction that hides low-level network details, CellOS automatically generates distributed solution programs to be run at individual base stations to collectively and collaboratively optimize the given network objective.

Key benefits
- Hides low-level network details through a virtual network abstraction so telco operators easily define centralized and high-level control objectives
- CellOS is independent of the underlying network and can be adapted to current and future cellular networks
Commercial Applications
- The technology has the potential to be implemented in next-generation cellular networks to simplify network management and improve performance in general
- Telco Operators can guarantee QOS while reducing the labor required to manage
Patent Information:
For Information, Contact:
Myron Kassaraba
Director of Commercialization
Northeastern University
Tommaso Melodia
Leonardo Bonati
Salvatore D'Oro
5G Networks
Automatic Decomposition
Cellular Networks
Distributed Optimization
Next-generation Mobile Networks
Software-defined Networks