CRAN - Campus Sciences
BP 70239 - 54506 VANDOEUVRE Cedex
Tél : +33 (0)3 72 74 52 90
cran-secretariat@univ-lorraine.fr
 
 
Ph. D. Project : Software Defined Network for greening internet communications
Dates : 2017/02/02 - 2020/01/23
Student: Md. Mohaimenul HOSSAIN
Manager(s) CRAN: Eric RONDEAU , Thierry DIVOUX
Full reference: General Context:
Today, network engineering is based on Service Level Agreement (SLA) signed between ICT users (customers) and ICT service providers. SLA must define both services and levels of services expected by ICT users. ICT service providers must design ICT physical architectures offering Quality of Services (QoS) adapted to ICT users’ requirements. ICT engineering approach is then mainly oriented on ICT physical infrastructure with the consequence to map each SLA to specific ICT physical solutions. A new paradigm named Software Defined Network (SDN) is emerging in network community consisting in a new software layer between ICT users’ services (business side) and ICT physical infrastructure. Main SDN objectives are to provide (1) agile solutions enables end-to-end deployment of service without any physical changes, (2) automation procedures to reduce errors and lead times, (3) ICT infrastructures more reliable with approaches based on reprovisioning and not repair on failure, (4) solutions better used with optimal placement of resources and in avoiding purchasing in advance of demand,… In term of business, Gartner claims that by 2017 more than half of all enterprises will have adopted an architectural approach similar to that of the cloud giants and Gartner sees software defined applications as a top technology trend for 2015. Moreover, International Data Corporation (August 2014) explained “The worldwide SDN market for the enterprise and cloud service provider segments will grow from $960 million in 2014 to over $8 billion by 2018”. In the same time the scientific community is developing new journals and conferences in SDN (see : http://sdn.ieee.org/).

Green Networking Context:
Globally, in 2008, network-enabled devices consumed about 420 TWh of electricity corresponding to the current electricity consumption of France. And the estimated global energy demand of network-enabled devices is expected to reach around 1140 TWh by 2025 (Bio Intelligence Service, 2013). The energy consumption has an impact on both OPEX for companies and on environment (greenhouse gas, resource depletion). In November 2015, European Commission published a draft report on Best Environmental Management Practice in the Telecommunications and ICT Services Sector to limit the ICT negative effects on earth. Nevertheless, new SDN approach is not considered in these best practices.

Aim:
The objective of this research is to make greener telecommunication and network infrastructure from SDN context in considering two aspects of sustainable development: (1) energy/Co2 emission and (2) resource management. The goal is to map physical ICT infrastructure with users’ services in identifying network clusters guaranteeing users’ service levels. From this identification, network clusters can be powered or slept in real-time regarding the service usage (proportional computing, selected connectedness). The advantage of cluster identification is also to plan maintenance operations on temporarily unused clusters. Moreover, network devices outside clusters can be defined as Zombie machines meaning the machine has no link with one business service. Zombie machines can be then switched off and dismantled to be recycled or refurbished for other applications. The PHD goal is split in two parts:
(1) The theoretical part consists in developing clustering algorithms based on graph theory. The results could be assessed using network simulation tool (such as Riverbed tool).
(2) The implementation part is the specification and development of new application services inside Openflow protocol. Openflow is the protocol used by SDN to control physical network.
Keywords: Network, Internet, Green networking, SDN
Department(s):
Eco-Technic systems engineering
Publications: hal-00953000, hal-00144780,hal-01145768,hal-01205751    + CRAN - Publications