Titulo Estágio
Intelligent monitoring of crítical communications services across IoT-Edge-Cloud compute continuum
Áreas de especialidade
Comunicações, Serviços e Infraestruturas
Sistemas Inteligentes
Local do Estágio
Instituto Pedro Nunes, Coimbra
Enquadramento
The increase in mobile traffic , stemming from the increased availability of connected devices (e.g., smartphones, tablets or IoT devices such as smart speakers) reinforces the need and interest in technologies such as Edge Computing, which enables resources and applications to be managed (e.g. processed) closer to the edge of the communications network (e.g. 4G, 5G) or the device itself. In other words, this processing is realized closer to the consumers of services, which leads to decreased transmission latency and traffic distribution across the network (vs its centralization in the Core), and ultimately to improvement in users satisfaction’ (Quality of Experience). The availability of Edge Computing promises to accelerate the transition to a new era of innovative services such as the Tactile or Sensorial Internet , Industry 4.0, connected vehicles or smart cities (maybe even the next “Killer App”?), which justifies moves both across the telecommunications sector , from Cloud Hyperscalers (e.g. AWS Outpost, Azure) or even industries such as manufacturing (e.g. Bosch IoT Edge).
This internship is aligned with P2020 OREOS (Orchestration and Resource optimization for rEliable and lOw-latency Services), a project led by Altran Portugal which aims at developing innovative solutions towards the support of critical wireless communications (i.e. requiring low latency and/or high reliability) in mobile scenarios by exploiting advances in Edge Computing, IoT, 5G networks, AI or Software. Critical wireless communications will be key for future services and advanced use cases such as Smart Cities, Intelligent Mobility or next-generation Public Protection and Disaster Recovery (PPDR) services.
1) Cisco Annual Internet Report (2018–2023) White Paper
2) “10 Hot Consumer Trends 2030”, Ericsson ConsumerLab, December 2019 https://www.etsi.org/newsroom/press-releases/1865-2021-01-etsi-announces-mec-sandbox-for-edge-app-developers
Objetivo
Given the computational and energy limitations of the terminals (traditional or IoT), and considering network (and resources overall) utilization may significantly vary over time, it is necessary to guarantee integrated monitoring and intelligent management across all of the devices (including IoT) and network resources (i.e. Edge and Cloud infrastructures). As a simple example, once a predefined threshold is triggered (e.g. average service delivery latency > 5 s), a given function or service (e.g. video decoding) may be moved dynamically from Cloud to Edge; for another trigger (e.g. device CPU usage > 70%) other functionalities may transition between the IoT device and the network. Considering strict Telecom regulation, it is crucial that such resource management is handled with no service downtime (targeting > 99,999% service availability).
The main objective of this internship is to study, design and implement intelligent infrastructure monitoring solutions allowing the collection of service and resources metrics from different domains, their processing and analysis; additionally, and based on aggregate / historical data, the solution should be able to learn and predict network or service events applying techniques such as time series forecasting and Machine Learning models (e.g. Regression, Random Forests).
Towards these goals, the internship will involve the study and analysis of open-source data streaming, analysis and processing technologies, and existing monitoring platforms suitable for highly virtualized 5G Telecommunications networks, and aligned to so-called Telco Cloud requirements (e.g. supporting Cloud Native applications).
Plano de Trabalhos - Semestre 1
The tasks to be performed during this work are described next. The student is expected to produce a report at the end of each phase with the description of the output of the tasks performed.
First Semester:
Phase 1 [Week 1 - 8] - Study of state-of-the art on network monitoring, specifications including ETSI’s Multi-Access Edge Computing and ETSI NFV specifications; analysis and experimentation monitoring tools (e.g. InfluxDB, Prometheus) and intelligent data analysis tools (e.g. Apache Druid, Drools), and network emulation tools like Mininet.
A small test emulation of a Data Center using a emulation environment such as Mininet may be created at this stage, enabling the experimentation of one or more of the selected tools.
Phase 2 [Week 8 - 16] - Monitoring solution requirements analysis, and baseline monitoring solution selection
Phase 3 [Week 16 - 20] - Writing of first semester report (UC delivery deadline)
Plano de Trabalhos - Semestre 2
Second Semester:
Phase 4 [Week 1 - 3] – Use case definition and test configuration Setup
Phase 5 [Week 3 - 9] – Implementation and integration of monitoring platform
Phase 6 [Week 9 - 15] - Exploitation / implementation of enhancements to the monitoring platform (e.g. targeting latency and reliability improvements, event forecasting features)
Phase 7 [Week 15 - 18] – Solution evaluation according to defined metrics
Phase 8 [Week 15 - 20] - Writing of final report (UC delivery deadline)
The dissertation will be supervised by Hugo Amaro (hamaro@ipn.pt)
Condições
Bolsa no valor de 400€ /mês durante um período de 6 meses, mediante avaliação do estagiário a ocorrer no fim do primeiro semestre. O valor pode ser revisto em alta aquando da avaliação.
O trabalho será realizado no Laboratório de Informática e Sistemas (LIS) do Instituto Pedro Nunes (IPN).
Observações
Risks and contingency:
The implementation of the enhancements to the monitoring platform include an AI component that predicts network or service events based on historical data. In case the datasets of historical data are not available or they are made available late relative to the timings of this internship, there will be a new development phase in the Second Semester, consisting in "Network events data generation and collection in a controlled simulated environment", prior to Phase 5.
Orientador
Hugo Dinis Pereirinha da Silva Amaro
hamaro@ipn.pt 📩