network management and analysis
functional connectivity inference in large-scale networks
We have proposed a novel approach for inferring functional connectivity in large-scale networks from time series of emitted events. This work is the first to introduce the concept of functional connectivity, prominent in neuroscience, to network management. We propose a data-driven inference approach that enables network operators to quickly identify network and service outages by calculating the statistical dependence between the times of events emitted by network devices.
Internet traffic analysis
In the context of traffic analysis, we have conducted one of the most comprehensive studies on Internet traffic in the literature, which is unique with respect to the spatial and temporal diversity of the studied Internet traces. Through rigorous statistical analysis of publicly available traces, we showed that the log-normal distribution is the best fit for Internet traffic volume. The work has major impact on designing models for researching networking problems, SLA provisioning and 95th-percentile pricing for ISPs. The work was published (reviewed in top 5% amongst 1464 papers) at IEEE INFOCOM.
vertex entropy for effective network monitoring
We have been working on identifying critical nodes in a network in order to eliminate network events emitted by non-critical nodes. We demonstrated that vertex entropy is effective for determining the significance of events generated by devices in large-scale computer networks (rigorously evaluated with data from a commercial network), which cannot be done with existing approaches. Vertex entropy is fully integrated in Moogsoft’s AIOps suite and deployed at Fortune1000 companies. We have also presented a novel way of describing the evolution of networks that is in better quantitative agreement with real world networks than the preferential attachment model.
funding
This work has been funded by EPSRC, Innovate UK and Moogsoft.
publications
G. Winchester, G. Parisis, and L. Berthouze, On the temporal behaviour of a large-scale microservice architecture, in Proc. of IEEE/IFIP AnNet, 2023.
G. Winchester, G. Parisis, and L. Berthouze, Accelerating Causal Inference Based RCA Using Prior Knowledge from Functional Connectivity Inference, in Proc. of CNSM, 2022.
A. Ibraheem, Z. Sheng, G. Parisis, D. Tian, In-Vehicle Network Delay Tomography, in Proc. of IEEE GLOBECOM, 2022.
G. Winchester, G. Parisis, and L. Berthouze, “Exploiting Functional Connectivity Inference for Efficient Root Cause Analysis”, in Proceedings of IEEE/IFIP NOMS, 2022.
M. Alasmar, R. Clegg, N. Zakhleniuk and G. Parisis, “Internet Traffic Volumes Are Not Gaussian – They Are Log-Normal: An 18-Year Longitudinal Study With Implications for Modelling and Prediction”, in IEEE/ACM Transactions on Networking, vol. 29, no. 3, pp. 1266-1279, 2021, doi: 10.1109/TNET.2021.3059542.
A. Ibraheem, Z. Sheng, G. Parisis, D. Tian, “Neural Network based Partial Tomography for In-Vehicle Network Monitoring”, In Proceedings of IEEE ICC 2021 Workshop on Time-sensitive and Deterministic Networking, 2021.
A. Messager, G. Parisis, I. Z. Kiss, R. Harper, P. Tee and L. Berthouze, “Inferring Functional Connectivity from Time-series of Events in Large Scale Network Deployments”, in IEEE Transactions on Network and Service Management, vol. 16, no. 3, 2019.
M. Alasmar, G. Parisis, R. Clegg, N. Zakhleniuk, “On the Distribution of Traffic Volumes in the Internet and its Implications”, In Proceedings of IEEE INFOCOM, 2019.
A. Messager, G. Parisis, I. Z. Kiss, R. Harper, P. Tee and L. Berthouze, “Functional Topology Inference from Network Events”, In Proceedings of IFIP/IEEE IM 2019.
P. Tee, G. Parisis, L. Berthouze and I. Wakeman, “Relating Vertex and Global Graph Entropy in Randomly Generated Graphs”, in Entropy 2018, 20(7), 481.
A. Messager, G. Parisis, R. Harper, P. Tee, I.Z. Kiss and L. Berthouze, “Network Events in a Large Commercial Network: What can we learn?”, In IFIP NOMS AnNet Workshop 2018.
P. Tee, I. Wakeman, G. Parisis and J. Dawes, I. Kiss, “Constraints and entropy in a model of network evolution”, in European Physical Journal B, 90 (11) 226, 2017.
P. Tee, G. Parisis and I. Wakeman, “Vertex Entropy as a Critical Node Measure in Network Monitoring”, in IEEE Transactions on Network and Service Management, vol. 14, no. 3, 2017.
P. Tee, G. Parisis, and I. Wakeman, Towards an Approximate Graph Entropy Measure for Identifying Incidents in Network Event Data, In Proceedings of the IEEE/IFIP International Workshop on Analytics for Network and Service Management (AnNet), 2016.