Middleware for safe human-robot collaboration

Initiated in 2021, MIRC builds an intelligent middleware based on deep learning and time-sensitive networking (TSN). MIRC focuses on risk collision in human-robot collaboration environments. Such collisions may cause physical damage and financial loss. It develops and evaluates a number of intelligent underlying tools for the development of applications that mitigate accidents. The proposed safety-oriented middleware is based on AI techniques and supports risk-related decision-making. It combines and orchestrates machine learning algorithms to provide a low-risk inference time. The project explores new techniques for the compression of deep neural networks in order to run these onto limited IoT devices. Also, it use a more reliable and latency aware communication network among components of the middleware. It runs on a TSN testbed for critical risks assessment. All experiments are conducted in a well-controlled scenario containing multiple people making collaborative tasks with a UR-5 robotic arm.

Scalable Service Function Chain Allocation 2 (SFC-DCA2)

Project started in 2021 with an industry partner. Its main goal is to propose a Service Function Chain (SFC) allocation solution that primarily focuses on large scale infrastructures. The study examines the scalable placement and the composition of VNFs in a typical distributed data center scenario. Given the limitations in terms of memory consumption to process an entire graph that represents the network infrastructure and computing resources, this work presents a proof-of-concept to demonstrate the scalable processing of large-scale resource graphs in memory. A 5G scenario for scalable SFC allocation is considered.

Benchmarking, VNF Optimization and Predicting SFC Performance

Starting in 2022 and in cooperation with the Sao Paulo research foundation FAPESP (www.fapesp.br), we look at building a framework that given a Linux server software and hardware configuration, it predicts the performance issues at the VNF´s dataplance. This will then also be extended to explore the performance of service function chains (SFCs).

Sentinella – Internet of Things for the Management of Water Resources

This project initiated in 2022 and is managed by APAC, a water management company at the state of Pernambuco and supported by the local state research foundation FACEPE (www.facepe.br). The research develops IoT devices and software for the monitoring, transmission and processing of state wide water reservoir levels and water quality indexes.

Automatic Generation of Stochastic Models to Estimate Next-Generation Cloud Data Center Availability

Research Productivity Grants from the CNPq research agency.  The general objective of this project is to expand the models, already developed previously, and integrate the Redfish toolchain with stochastic models and algorithms, supporting features such as automatic acquisition of IT subsystem configuration to estimate data center and Virtual Performance-Optimized Datacenters (vPOD) availability.