1. Fabrizio Granelli (granelli@disi.unitn.it)
FP7 PACE Project Workshop on "Path Computation Element and beyond: innovating control and management functions towards fully reconfigurable software-centric networks"
Software defined and virtualized wireless access open issues and perspectives
Fabrizio Granelli

2. Table of Contents Introduction Extending SDN over wireless E2E SDN
Virtualizing wireless access
NFV
Test case: SDN for interference control
E2E SDN
Standardization
Conclusions

3. Introduction
Computer networks play a very important role in our day to day life
As the number of users increases so does the number of network devices
Challenges:
Configuration and management becomes costly
System updates are very demanding
Lowers the pace of innovation

4. Introduction (cont’d)
Why?
The control logic (intelligence) is embedded to every device
How it could be solved?
Software-Defined Networking (SDN)
What is SDN?
Newly emerging network architecture paradigm
Promises innovation in terms of:
Network programmability
High-level of abstraction

5. Introduction (cont’d)
How is it achieved?
By separating the control-plane and data-plane with a well-defined programmable interface to provide:
Centralized global view of the network
Easy configurability and manageability
Flexibility, scalability and capability of reconfiguration of different network segments

6. Introduction (cont’d)

7. Reasons for Mobile Traffic Boom
Why wireless?
Mobile data traffic is also booming due to:
Reasons for Mobile Traffic Boom

8. Why wireless?

9. Increase in capacity demand Mobile data traffic boom
Why wireless? (cont’d)
Device bandwidth consumption profile*
Growth in the number of devices  growth in mobile data traffic
Device
Traffic
M2M
6 *
Wearable Devices
7 *
Smartphones
49 *
Tablets
127 *
Laptops
227 *
-CAPEX/OPEX
Densification:
-Interference
- Spatial reuse
-etc.
Increase in capacity demand
Mobile data traffic boom
To account these issues, programmability of mobile networks is needed:
Allocating non-overlapping frequency bands
Monitoring interference
Coordinating handovers
Configuration and management
*Cisco and/or its affiliates, ”Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2013–2018”, White Paper, February 5, 2014

10. Extending SDN to wireless
Extension of the NaaS concept to include wireless

11. Google Project Fi

12. Wireless access virtualization perspectives
Flow Oriented Perspective
Also, mobile network virtualization
Management and scheduling of different flows on different slices
Implemented as overlay (OpenRoads, vBTS) or via internal scheduler (NVS, vLTE)
Protocol Oriented Perspective
Customize and manage multiple wireless protocol instances on the same radio HW (OpenRadio, SORA)
Spectrum Oriented Perspective
RF bands and raw spectrum are sliced

13. Wireless virtualization
Extension of Service Awareness and Functions Modularity to the wireless domain
Decoupling of management and data planes (CAPWAP)
Virtualization using SDRs
Baseband processing divided into programmable processing and decision planes
WLAN Virtualization
Connecting NICs to different networks
CloudMAC
Cellular Base Station Virtualization
Different schedulers for different slices
Wireless Spectrum Virtualization
Spectrum reshaping (Spectrum Virtualization Layer)

14. Enabling SDN over wireless
The idea is to augment the capabilities of WiFi (Virtual WiFi)

15. Enabling SDN over wireless (cont’d)
Efficient management and allocation of network resources (NetShare)

16. Enabling SDN over wireless (cont’d)
- Additional flexibility and efficiency in resource allocation

17. Enabling SDN over wireless (cont’d)
CROWD architecture (
Extremely dense and heterogeneous scenario
Mitigation of intra-system interference, opportunistic channel usage, energy efficiency
Control through CROWD Regional Controller and CROWD Local Controller

18. Enabling SDN over wireless (cont’d)
Enhancing IP multicast over networks
Multiflow converts multicast tx into unicast, to optimize channel availability

19. Test case: SDN for Interference control
Even though the problem of interference was studied for decades at both PHY and MAC layers, so far no satisfactory solution emerged.
Objective:
To come up with a framework for interference control (i.e., manage and avoid) to mobile networks by adopting the SDN paradigm
(*) Anteneh A. Gebremariam; Leonardo Goratti; Roberto Riggioy; Tinku Tasheed; Fabrizio Granelli, “A Framework for Interference Control in Software-Defined Mobile Radio Networks,” IEEE CCNC 2015.

20. SDN for Interference control (cont’d)
How?
By exposing/abstracting the lower layers of the protocol stack (i.e., PHY and MAC) in order to make system parameters available to the controller or applications
Providing an efficient resource scheduling scheme to control interference
Interference control framework
The Goal is the design and implementation of the interference control framework
System Architecture

21. SDN for Interference control (cont’d)
The selection of H is done
by the central controller based on the channel feedback
information of the UEs
Y can be represented by a
tuple of parameters, <time, frequency, space, transmit-power, modulation, coding, antenna-port, beam-pattern>, representing an abstraction of the resources assigned to a data stream X
PORT
The details of each block in the interference control framework
PORT :- motivated by the switch port concept, where the switch input ports are mapped in to an output port based on the forwarding table.
In a similar manner we can use the same concept in the wireless domain to provide the flexibility of selecting the mapping H from the centralized controller.
Interference Graph (IG) :- models the interference among communication links. Its represented by two tuples (V, E); where V represents the UEs and E represents the edge connecting the interfering transmitter to the receiver (i.e., UE). From the figure above, the purple lines represent the IG and the black solid lines represent the connectivity graph.
IG construction:
I, By collecting information for a certain period from all communication links
ii, Micro-probing: by injecting traffic to the network (accurate and very demanding)
Interference Graph (IG)
IG represented by a pair (V, E)
IG construction:
Based on the history
Micro-probing

22. SDN for Interference control (cont’d)
Conflict Graph construction
What are weights assigned for each edge?
The maximum allowed interference level at
the receiver (using Physical Interference Model) is:
The maximum allowable interference contribution
of the kth interfering link on terminal XR(i):
Conflict Graph (CG)
Where β is the threshold SINR, σ2 is the additive noise, Xi and XR(i) represents the transmitter and receiver location
Sm represents a set of communication links that could be active at the same time, considering the following expression is satisfied:

23. SDN for Interference control (cont’d)
The optimizer:
Goal: to reduce the weights in a certain transmission link in the CG
How?
Efficient resource scheduling: Integer Linear Problem (ILP)
Objective function – minimization of the weights (i.e., minimizing the aggregate interference)
Where: L- represents the transmission links (UE, eNB) pairs
R- represents the resource blocks (RBs) (frequency, time, space)
M-the Modulation Coding Scheme
Φi,k,r,m :- is a decision binary variable, which is 1 if link li uses MCS m in RB r or 0 otherwise
the power constraint sets the interval for the transmission power level
TP constraint makes sure that each link achieves its throughput demands 
Constraints :
decision binary variable, which is 1 if link li uses MCS (modulation and coding scheme) m in RB (resource block) r or 0 otherwise

24. End-to-End SDN
To enable combined management of wireless and wired segments of the network
Advantages:
Unified management of the network
Unified policy enforcement
Enabler for NaaS
Performance improvement
Vendor interoperability
Customized applications through SDN (northbound) APIs

25. Current Standardization Efforts on SDN
IETF Forwarding and Control Element Separation (ForCES) WG
Open Network Foundation
Pushing OpenFlow
Interfaces between (1) applications and controller and (2) controller and switching infrastructure
ITU-T SG13 (Future Networks) and SG11 (SDN signaling)
IRTF Software Defined Networking Research Group
ETSI SDN/NFV

26. Standardization of wireless NVFs
Several interesting scenarios of wireless in the future
5G calls for heterogeneous wireless networking
SDN could come in the picture to enable efficient management of 5G infrastructures
Standardization could be two-fold:
Standardizing WHAT is virtualized
Standardizing the WAY we compose virtualized entities

27. Conclusions
SDN is expected to be “hot” both in wireless as well as in wireless-cum-wired scenarios
5G could represent a suitable scenario, including hetnets and softwarization
Different opportunities for standardization
For more info: F. Granelli, et al. “Software Defined and Virtualized Wireless Access in Future Wireless Networks: Scenarios and Standards”, IEEE Communications Magazine, in press

28. Any questions?
Software defined and virtualized wireless access open issues and perspectives
Fabrizio Granelli