OpenLab

OpenLab brings together the essential ingredients for an open, general purpose, and sustainable large-scale shared Future Internet Research and Experimentation (FIRE) Facility. We advance early prototypes of this Facility. These are testbeds, middleware, and measurement tools that we extend to provide more efficient and flexible support for a diverse set of experimental applications and protocols. The prototypes, coming from former FIRE initiatives OneLab and Panlab, as well as other valuable sources, include a set of demonstrably successful testbeds: PlanetLab Europe, with its 150 partner/user institutions across Europe; the NITOS and w-iLab.t wireless testbeds; two IMS telco testbeds for exploring merged media distribution; the GSN green networking testbed; the ETOMIC high precision network measurement testbed; and the HEN emulation testbed. Associated with these testbeds are similarly successful control- and experimental-plane software. OpenLab advances these prototypes with key enhancements in the areas of mobility, wireless, monitoring, domain interconnections, and the integration of new technologies such as OpenFlow. These enhancements will be transparent to existing users of each testbed, while opening up a diversity of new experiments that users can perform, from wired and wireless media distribution to distributed and autonomous management of new social interactions and localized services, going far beyond what can be tested on the current Internet. OpenLab‘s interoperability work will bring FIRE closer to the goal of a unified Facility and will provide models that we will promote to the Future Internet PPP. Finally, we will, through open calls, support users in industry and academia, notably those in FP7 Future Internet projects, who propose innovative experiments using the OpenLab technologies and testbeds.

Start Date: 01/09/2011

Duration: 30 months

Openlab WP3 (UTH leads this WP):

This workpackage focuses on the enhancement of a number of existing wireless testbeds with a specific focus on supporting mobility related experiments. Enhancements will be in the following areas:

• Add new capabilities, such as support for LTE
• Extensions to the resource description ontology addressing the wireless and mobility domain
• Extend existing user tools to incorporate the above resource description
• Extend NITOS scheduler for a federated environment with  ̳pluggable‘ policies
• Extend support for mobility focused experiments, specifically through: i) placing testbed resources on robots whose mobility can be controlled by the experimenter, ii) programmatic control of the wireless attenuation between devices to emulate mobility and iii) support for disconnected operation from the control plane for devices carried by users
• Extensions to the instrumentation and measurement framework to: i) capture device location indoors as well as outdoors and ii) support collection and management of privacy-sensitive data for experiments with real users
• The testbeds concerned by this workpackage include NITOS, DOTSEL and w-iLab.t

NITOS Hardware Extensions within the context of OpenLab:

NITOS will be extended with:

• LTE Base Stations
• 3G-femtocell Base Stations

 Project's website: http://www.ict-openlab.eu/

Cooperative Networking for High Capacity Transport Architectures (CONECT)

CONECT proposes a holistic network design approach that drastically enhances performance in wireless networks by unlocking the hidden potential of the broadcast wireless medium. Information and communication theory, and protocol design engineering are used as the foundations for developing provably optimal cooperative information forwarding strategies, from multi-signal fusion to packet relay and node reciprocation. As a result, close to optimal end-to-end throughput is achieved while energy consumption is significantly reduced and other performance metrics like delay can be controlled.

Wireless information transport is currently realized under the highly suboptimal assumptions of point-topoint communication and strict signal separation. Recent network information theory advances suggest approaches for constructive exploitation of the multiplicity of signals arriving at a wireless node, traditionally considered as harmful interference. Building on this foundation, CONECT will develop an architecture that exploits these principles all the way from the physical communication substrate to cooperative packet level access and forwarding, to application level information exchange. Many-tomany information exchange modes like multicasting and any-casting will particularly benefit from this approach as the broadcast effect inherently amplifies the multicast transport capability. This constitutes a significant advance from more traditional approaches where interference is treated as a foe.

A concrete experimentation plan will validate the design choices and at the same time provide real-world feedback for fine-tuning of the architecture. It will be based on the OPENAIRINTERFACE test-bed for flexible radio network design and the NITOS test-bed for cooperative wireless access and transport. A video multicast application will provide the basic evaluation scenario. NITOS, being an ONELAB testbed, will provide the link to federation with the Planetlab-Europe infrastructure.

Project's website: http://www.conect-ict.eu/

Motivation: Network coding may not be beneficial in multi-rate settings

In settings where different transmission rates are employed to combat interference, network coding may not always provide the favorable performance. Rate control is a traditional, effective means of coping with interference in wireless networks: the transmitter adapts the bit rate as per the quality of the link with the receiver. For example SampleRate selects the bit rate with the highest observed link throughput; the lower the throughput, the lower the bit rate. With regards to network coding this is translated to transmitting encoded packets at the lowest bit rate that can be supported by all intended receivers, in order for any of them to be able to receive those encoded packets. As an example, consider the scenario in the following figure, where nodes (1) and (3) wish to exchange packets (A) and (B) through the relay node (2), and f(a, b) is the highest-throughput bit rate on a link a → b.

While 4 transmissions are traditionally required, with network coding the packet exchange can take place in 3 transmissions, since node (2) is transmitting a decodable linear function (A+B) of the packets to nodes (1) and (3). However, even though the same amount of information is exchanged with fewer transmissions, network coding will not lead to throughput improvement, when:

• f(2,1) « f(2,3). In this case, the relay node (2) will have to transmit the packet (A+B) at rate f(2,1); otherwise node (1) will not be able to decipher the packet. The use of this rate can be suboptimal for the link (2)→(3).

• f(1,2) « f(3,2), while f(1,2) « f(2,1). In this scenario, node (2) will have to wait for a long time until the reception of packet (A) from node (1), since the rate f(1,2) is very low. This will also delay the transmission of packet (A+B); hence the throughput on link (2)→(1) will not be significantly incresed.