Details

Hits: 491

IoT Lab Experimental Project

Cognitive and User-centric Networking Area

TEAM:

• Rute C. Sofia (PI)
• Liliana I. Carvalho, NEMPS PhD Student/Jr Researcher
• Daniel Maniglia Silva, NEMPS PhD Student/Jr Researcher
• Helder Valente, NEMPS PhD Student
• F. Melo Pereira, PhD Student
• Godwin Asamooning, PhD Student
• Hector Orrillo, PhD Student
• José Soares, MSc Student ULisboa
• José Ardions, MSc Student Universidade Lusófona
• André Vaz, MSc Student Universidade Lusófona
• Roberto Jr, MSc Student Universidade Lusófona
• Rogério Costa, MSc Student Universidade Lusófona
• Former Team Members

Description

 The IoT Lab is an operational team of COPELABS that works on networking architectures and communication protocols for multiple IoT environments, e.g., personal IoT and industrial IoT. A few aspects being worked upon concern protocolar transmission and unified abstraction models for edge computing, as well as the relevancy of novel communication models (e.g., ICN) for IoT.

Internet of Things (IoT) communication architectures and protocols are evolving to be able to cope with new challenges such as the processing of large amounts of data; filtering; data mining and classification; high heterogeneity in devices and software.

The evolutionary trend and interoperability aspects of publish/subscriber approaches as well as of network architectures and protocols in general are being exploited by the COPELABS IoT Lab team.

Current Activities

Performance Evaluation of Scalable Networking Architectures for IoT (J. Soares, , R. C. Sofia, P. Veiga (ULisboa))

In this line of work the lab is evaluating different IoT communication protocols for IoT scenarios (industrial and consumer) in terms of packet loss, latency, deployment easiness, memory and CPU use.

Exploring the role of context-awareness on the network to improve data transmission and the network operation in personal IoT scenarios (D. Maniglia Silva, R. C. Sofia)

The motivation to develop this work concerns the belief that Personal IoT (PIoT) requires a different, more interactive type of communication to scale well in heterogeneous, mobile environments. Such communication includes feedback close-to-realtime to users.

This work is formulated around the following research questions:

Distributed edge computing to support large-scale pervasive sensing frameworks (L. Carvalho, R. C. Sofia)

Nowadays, mobile phones’ usage is increasing as an integral part of the daily lives of billions of people, being carried everywhere and used as part of the most varied aspects of our daily routines [1]. In this context, there has been recently a surge of related work focused on the development of middleware and subsequent studies that can assist, in a less intrusive way, a better understanding of different dimensions of well-being, by combining activity recognition based on fused data capture derived from multiple sensors with computational proposals for statistical inference of different behavioral aspects, e.g., sociability measures; mood evolution over time, etc. Despite the fact that there seems to be a growing interest in the aforementioned context, today there is no clear understand on how to best develop such tools, which sensors are best applicable to which type of activity detection and/or recognition; which models best suit learning and inference, where and when to capture data, how, when, and where to treat and to classify captured data .

In this work, we intend to develop a non-intrusive structure for inference of social interaction as well as of activities relevant in the context of social interaction (e.g., sleep) based on personal devices. The aim is to develop a system to be used in mobile phones or another user device in a distributed and secure way, where individual data is collected non-intrusively during the daily routine of people. We envision such framework to have at least modules concerning data capture; learning; inference; feedback (individual and collective). Hence, our contributions are four-fold: i) to provide a better analysis on current sensors available, and how their captured data can be combined to perform different types of activity recognition and behavior; ii) to perform validation (based on living labs) of the potential for such tools to stimulate social interaction thus, for instance, improving communication and giving rise to higher levels of sociability; iii) to come up with new algorithms, heuristics, mechanisms that can assist in further developing sensing middleware in a way that allows pervasive sensing analysis to scale to large sets of users in close-to-real-time; iv) to validate our findings by building a framework (proof-of-concept) that integrates our contributions.

Exploiting the role of smart data for location tracking (Points of Interest Detection via smartphone wireless sensed data), F. M. Pereira, R. C. Sofia

Cities present increasingly surprising aspects regarding Points of Interest (POI). For instance, everyday POIs are created; however, the majority of citizens may not even become aware of them. The problem, in our opinion, lies in multiple factors. Firstly, there is the unawareness of the interest for the populations in certain points of their city, i.e., there seems to occur a mismatch between created POIs, and the daily lives of citizens.

Secondly, the interests of all existing points of natural form, like beaches, landscapes or any natural resource, or created by public or private entities, such as gyms, entertainment areas, office buildings, or commercial areas and services, are not shared by all entities in the same (unified) way. Some information is not even permanent. This therefore difficult the understanding of the relevancy of specific POIs, as well as makes it harder to understand the interest that citizens have in specific POIs.

Thirdly, the main points of interest for the populations have to do with the explicit definition of the identity of the city. In addition to the POI's inherent to the identity of the city, which the managers of the city will advertise by its own way residents and visitors, there are POI's not advertised but with certain general interest to those populations and that are not listed in pre-established catalogues. These points are the hardest to detect, as they are usually unreferenced or just because they belong to thematic circles of interest. Imagine a Bullfight Plaza, it can be considered of restricted interest to followers of bullfighting. However, seasonally such a POI may attract large masses of visitors to witness an event that can bring together up to several thousands of spectators, residents and visitors.

In the core idea of this work is the development of a computational engine capable of assisting the detection of PoIs derived from wireless network mining.

Energy-efficient location tracking via personal IoT devices (A. Vaz, R. C. Sofia)

IoT technology is today scattered around and being used in the most diverse areas of one’s living. While specific hardware solutions, such as RFID, can be applied on a specific location, today the most versatile means of support for location tracking are: personal devices, often equipped with short range wireless technology, such as Bluetooth, Bluetooth Low Energy, or Wi-Fi.

On the other hand, position detection and location tracking are services that are today integrated into multiple aspects of technology, usually based on GPS. Nevertheless, GPS does not reach all places; does not perform well indoors, and is expensive in terms of battery.

Several vendors, such as Google, provide libraries to assist in location tracking via sensor fusion, i.e., via relying on GPS as well as on the location of Access Points in wireless networks.

However, short-range wireless technology available today can be used to provide quickly location tracking in a way that assists the user privacy. For instance, assuming that one is in a shopping-mall garage, looking for a car, then BLE or Wi-Fi Direct can be used to look for the object in a simple, low energy consumption way.

Similarly, if one is walking a pet, and the pet runs out of sight, short-range wireless technology can quickly assist in finding out the whereabouts of the pet.

This master thesis is proposed in this context, with the aim of contributing two-fold. Firstly, to implement an open-source concept of a location tracking embedded device, associated with the mobile phone. Secondly, to contribute to the performance evaluation of formulas that assist in a more efficient location tracking solution being the KPIs for more efficiency: lower energy consumption; faster detection of the object (lower latency).

Context-aware Edge Computing for Self-Driving Cars (Roberto Junior, R. C. Sofia)

In the context of the Lab, this specific Master Thesis concerns the analysis, experimentation, and validation of different edge computing solutions to the use-case of self-driving cars. Self-driving vehicles, in contrast to automated vehicles, do not require human intervention and can be seen as information silos with particular features, such as mobility patterns. Self-driving cars are in this work an example of an IoT device with mobility, sensing, and that is always connected to multiple networks, and devices. Self-driving vehicles send critical contextual data, such as weather changes, road conditions, or even driver conditions to other vehicles, and for that purpose produce and consume large volumes of data (0.75 to 1 GB data per second) to make real-time decisions for accurate navigation¹.

Today, such data is sent to the cloud to be processed, thus increasing latency and preventing a faster system reaction. These devices can, however, benefit from edge computing techniques, as edge computing² assists in moving computing functions, classification and learning closer to the data source(s).

The context of this dissertation is the exploration of contextual-awareness edge based solutions to offload time-sensitive data, and to compare performance with common cloud based solutions, for the specific case of self-driving vehicles. For that purpose, this work shall focus on a follow-up of the open-source Contextual Manager solution [6], adapting it to the needs of the specific use-case of self-driving cars. This implies, for instance, acquiring new indicators (e.g., car speed) and to revise the proposed availability and centrality weights provided for the surrounding neighbors, to select “edges” to pass the information to. Offloading and context-based edge selection are the core of this dissertation, for the specific sub-scenario of IoT environments integrating self-driving vehicles.

1According to Intel CEO Brian Krzanich, an autonomous car can use up to 4,000 GB of data in just one hour of driving.

Communication Platforms for Emergency Scenarios (José Ardions, R. C. Sofia)

In emergency scenarios, the communication infrastructure often experiences partial or complete damages, undermining not just global communication, as well as local (people-to-people) communication. Nevertheless, opportunistic wireless routing and mobile crowd sensing can assist people-to-people communication even in the verge of intermittent access.

This line of work is focused on exploring direct device-to-device communication based on i) an ICN approach; ii) an IoT IP-based messaging approach (e.g., AMQP).

An Information-centric Smart Home Monitoring Kit (Rogério Costa, R. C. Sofia)

Today there are several vendor-based smart home applications which are being applied into controlling and automating our homes. Often, smart home kits involve simple operations e.g., sensor data capture, and action by actuators.

The communication between the different elements of a smart home kit is today based on different protocols, being often the choice IP-based communication, due to the interoperability advantages provided by IP in heterogeneous environments.

More recently, information-centric paradigms such as the Named Data Networking (NDN) architecture, are being applied in the context of Internet of Things (IoT) environments. NDN implements a receiver-driven publish/subscribe communication model based on stateful forwarding. As NDN is focused on information and not on the hosts (machines), its architecture brings in relevant features for IoT.

This master thesis is proposed in this context, with the aim of contributing to a better understanding of potential performance advantages that NDN may have in comparison to IP, for a specific scenario of Smart Home monitoring and control.

Wireless Networking for Autonomous Mobile Smart Cameras (Godwin Anuork Asaamoning, P. Mendes)

Related Projects:

Unified Communications in IoT (2017/2019)

BEING - Inference of Human Behavior via Networking (2018/2019)

Proxemics Data Lab (2016-2018)

Related Material:

- Sofia, Rute C.; Mendes, Paulo. An Overview on Push-based Communication Models for Information-Centric Networking. 2019

- D. Silva, FIT IoT Experiment – work developed in the context of the “Pervasive Communication Systems Course”, second semeste. 2018

- R. C. Sofia, an Overview of the COPELABS IoT Lab