The Nokia Bell Labs Israel team leads research in the field of distributed cloud networking. Here is a sampling of our recent projects.
Agile, reliable, maintainable software-based services
As networking services evolve from hardware-centric to software-based ones, we anticipate that the market success of these new solutions will be largely determined by the cost, reliability and operational simplicity of these services. The shift toward agile, reliable and easy to maintain software-based services calls for a new model where operators and network equipment vendors work in close collaboration. We call our service Dev-for-Ops and envision a development model where the services are developed by a vendor with a deep understanding of the operational environment and its life cycle management requirements. Our researchers are developing solutions that improve the time to market and quality of each step of the process.
Continuous Defect Prediction
Reducing errors by predicting software flaws
Nokia Bell Labs values quality and we aim to improve the quality of software products developed using DevOps and Agile methodologies. With our Continuous Defect Prediction project, we analyze a project’s defect history and predict which changes of the most recent software release are likely to cause defects. The novelty of the technique lies in the analysis of two consecutive releases of a software product. While other defect prediction techniques predict flaws on the entire code base, our strength lies in our ability to pinpoint areas that have changed and are likely to contain defects.
Repeated Combinatorial Test Design (CTD)
Effective testing and improved planning with CTD
CTD starts with pairwise testing methods for test planning and aims to improve test coverage at each test iteration. Our technique presents the ability to guide QA teams to plan increasingly effective tests such that higher-level interactions between tested parameters are covered over time.
Descriptors for Networking Cloud Applications
A static-analysis toolkit for cloud developers
We are developing technologies and tools to help cloud developers orchestrate Virtual Network Function (VNF). In addition to basic developments tools, our research focuses on creating a static-analysis toolkit that enables cloud developers to statically reason about VNF services and their chaining. We analyze possible conflicts in service description, or synthesize a specifically desired behavior. The project extends the Heat/Murano/Mistral OpenStack projects.
Behavioral Anomalies Detection
Removing errors with log analysis
We use log analysis to detect software and system anomalies. We can extract behavioral activity from the log and compare with models representing normal and abnormal behavior. Our project leverages known techniques and tools such as kTails and Perfume to gain insights into the behavior of a system from its log. We use a new comparison method to emphasize pieces of logs that expose problematic behavior. Our technique reduces the time to pinpoint the root cause of a problem.
Analysis for Network Function Virtualization (NFV)
Locating performance anomolies using Big Data
We are developing an analysis framework for NFV infrastructure. Our framework will use big data techniques (such as machine learning) to identify root causes of performance issues within NFV environment. The uniqueness of the approach lies in its ability to correlate system alarms and performance metrics across multiple layers of the infrastructure to reveal dependencies between alarms and anomalies.
Virtual Switches Performance Evaluation
Including Open vSwitch in NFV architecture
We are developing a model that describes the performance and performance implications of using Open vSwitch (OVS) as part of future NFV architecture. We are exploring the performance properties of OVS by systematically conducting measurements of different configurations. We can evaluate OVS in terms of bandwidth, latency and CPU usage and expose system bottlenecks and pitfalls.
Understanding multidimensional network resources
We study the placement of the VMs (Virtual Machines) that implement the network functions over the physical infrastructure. We can identify the success factors focusing on multidimensionality of the resources and on service chaining.
Security innovations for Cloudband CBMS
In collaboration with CloudBand and Team8 teams in Israel, and Nokia Bell Labs Villarceaux our researchers work on improving the security of CloudBand CBMS and identifying promising areas of Nokia Bell Labs Security innovations.