ProjectTime Critical Clouds

Competence Areas at RISE ICT
Time Critical Clouds

Time Critical Clouds (TCC)

Advances towards programmable, software-defined and virtualized network services call for novel mechanisms for traffic management across access and core network infrastructures as well as data centers, with capabilities of guaranteeing reliable connectivity and services in line with the application requirements. 

The objective of the project is to enable latency-bounded key-value requests for clients using applications deployed in geo-distributed data centers. This is a typical cloud setup, e.g., for large-scale intelligent transportation systems in the future.

The project is supported by Scania, Ericsson and Telia in the industrial advisory board.

Relevance to the next generation networks

Network technologies and methods based on information and autonomous decision making is fundamental for running highly distributed systems as fast as needed to increase traffic throughput while maintaining robust and reliable connectivity. Guaranteed end-to-end latencies from, for example, an autonomous vehicle and a data center providing recent traffic situation updates, is really a matter of added safety and central for a pleasant user experience.

Ongoing activities

We research and develop fundamental mechanisms for resource coordination and intelligent traffic forwarding in wireless and wired networks fulfilling specified requirements on service performance.

TCC is funded by the Swedish Foundation for Strategic Research (SSF) and is carried out in close collaboration between the Network Intelligence group (DNA lab) at RISE SICS and the Network Systems Laboratory (NSLAB) at KTH.

The Time-Critical Clouds testbed

The functionality of the testbed was recently demonstrated at the RISE ICT Open House in Kista 2018. Our testbed in TCC is unique, as it is the first hybrid-infrastructure (LAN/WAN/Cloud) in the world running state-of-the art methods for forwarding traffic over both wireless and wired technology all the way to the data center and back to the client. The infrastructure is partially programmable - we have implemented customized high-performance network and controller software in all network entities we control.Remotely controlled vehicle over the cloud

Although the testbed is an early prototype of what is to be continuously extended and developed over the next years (within and beyond TCC), we have already now shown that we can remotely control a customised robot vehicle, by forwarding control traffic from RISE SICS and KTH, to Amazon clouds in Europe (Frankfurt and London) for simplified processing (writes and reads) and back - and still run the vehicle smoothly.

This is indeed an encouraging result - running a vehicle remotely without full control of the infrastructure is difficult due to the countless sources of bottlenecks and delays. With methods from KTH and SICS (Metron/SNF, Kurma and AquaMet) we successfully handle some of the bottlenecks that normally arise from network function virtualisation, data center overload and varying radio conditions. 

Overview of the testbed setup.

RISE ICT Open House visitors were given the opportunity to test the performance of the current testbed with the objective of controlling the vehicle to push mini-footballs into a goal.  Quick reaction to end-to-end messages between the vehicle and the hand-held controller to achieve the necessary precision to score leaves no room for delays. The following movie clip shows the remotely controlled vehicle in action: 

Next steps

This is a first test deployment of integrated approaches towards key-value requests with bounded latency that will give us valuable insights in order to address various bottlenecks in wired and wireless network environments. The current testbed is under continuous development and will be used for measurements, testing and evaluation over the next years. 

Recent publications

External homepage: