Cognitive and User-centric networking

2018-2019 Main Work Topics:

• Unified communications in IoT (R. Sofia, D. Maniglia, H. Valente, L. I. Carvalho, J. Soares, J. Ardions). Aspects concerning protocolar transmission and unified abstraction models, as well as the relevancy of novel communication models (e.g., ICN) for both personal IoT (D. Maniglia) and Industrial IoT. This line of action involves cooperation with external entities.

• Pervasive Wireless Communications (P. Mendes, H. Orrillo, A. Goodwin). This line of work concerns novel paradigms to orchestrate autonomic computational functions within a distributed networked system. Aspects being worked upon relate with Distributed Networks, Fog Computing and Cognitive Networking; For instance mobile edge computing in regards to optimizing distributed networked systems such as smart camera networks, as well as autonomous drones and cars.

• Self-organization aspects in distributed sensing devices (S. Tomic, M. Beko). The work being developed concerns mechanisms to assist a system to react locally to the variations within sensing records (OSI Layer 1); hence, accurate determination of the sensor location where the deviations arise is key. By exploring the synergies between computational and physical components (J. Canto), one can form smart environments offering improved safety and efficiency in everyday life, e.g., smart parking; assistance for elderly or people with disabilities; monitoring of storage conditions and goods.

LONG-TERM PURPOSE:

We envision mobile systems to evolve to support a large set of data-centric networking services by including deeper awareness of mobile crowd behavior and context. We intend therefore to contribute to a broader notion of distributed cognitive networking, one that integrates the concept of programmable networks, allowing networked devices to perform customized computation closer to the end-user (user-centric networking). We envision that such generation of cognitive networks will rely on intelligence across all OSI Layers, raising the level of cognitive abilities (known as “Self-X” in the context of network management) via machine learning, e.g., to devise novel solutions related to adaptation and anticipation properties of the network elements across multiple OSI Layers.Since the Internet has evolved to be dominated by content sharing and retrieval, the opportunistic networking concept already under work in COPELABS is going to be exploited to handle data as a network primitive, decoupling data location from identity, security, and access, and allowing data to be shared directly among users.

This line of confluence shall also investigate open issues in the context of unified communication in IoT.

Another aspect of relevance for the success of IoT is its capability of setting up and controlling a communication layer among a set of distributed sensing devices. In such distributed communication systems, cooperation and localization is a crucial factor  since a system may be configured to react locally to the variations within sensing records; hence, accurate determination of the sensor location where the deviations arise is key. By exploring the synergies between computational and physical components one can form smart environments offering improved safety and efficiency in everyday life, e.g., smart parking; assistance for elderly or people with disabilities; monitoring of storage conditions and goods.

Computational Systems                              

2018-2019 Main Work Topics:

• Personal cloud systems (J. Faísca, J. Rogado). Continuing the activity under development since 2012 (J. Faísca, J. Rogado), initially exploring identity management and data protection aspects, this line of work evolved into exploring Distributed Ledger Technology (DLT), commonly referred to as Blockchain, as a way to deal with some of the requirements implied by this sparse connectivity environments.

• Fault-tolerance in cloud computing (P. Costa). Fault-tolerance is a major issue in the context of cloud computing, and it is expected to increase due to the development of cloudlets. As such, the challenge of building dependable and robust clouds and solutions remain a critical research problem. This topic focuses on analyzing the viability of combining the distributed consensus of Blockchain with MapReduce computation to sustain distributed cloudlet platforms in a way that they are tolerant to arbitrary faults, malicious faults and cloud outages, and still guarantee performance at an acceptable cost. During 2018 this area shall be explored via the development of a prototype that will combine the Blockchain paradigm with MapReduce computation to validate a real use case scenario.

 LONG-TERM PURPOSE:

 The computational systems research line intends to continue the focus on exploring decentralization technologies, in the context of next generation networks, i.e., cloudlet based mobile approaches. In addition to mobile storage and caching issues, decentralized solutions require mobile devices to be connected to their peers or to their cloud service providers, usually via cellular technology often with intermittent connectivity, which can be extremely limiting both in terms of cost, latency, as well as in terms of energy consumption. Continuing the activity under development since 2012 initially exploring identity management and data protection aspects, this line of work evolved into exploring Distributed Ledger Technology (DLT), commonly referred to as Blockchain, as a way to deal with some of the requirements implied by this sparse connectivity environments. Blockchain, a concept initially associated to the support of a digital currency (i.e.: Bitcoin), is currently generalized to many areas of activity. Its decentralized mode of operation, the immutable characteristics of its storage and the secure nature of its transactions, fit into several scenarios where the primary challenge relies on a secure and immutable transactional infrastructure connecting entities (humans or machines). These entities can exchange values, services, but also negotiate the controlled disclosure of personal data, such as identity claims or personal sensor information. The promising results obtained in recent academic work, which applied DLT to Semantic Identity Management and leveraged Ethereum smart contracts to negotiate sensor data pave the way for a future sustained line of research activity in the context of these emerging areas.

Data Mining & Statistics                                                                                 

2018-2019 Main Work Topics:

• Network mining (R. Sofia, L. Carvalho, F. Melo Pereira). A relevant aspect to address in the context of a next generation internet concerns data mining and analytics applied locally (e.g. in decentralized systems, or in cloudlets. In this context, eager classification models such as neural networks have been the preferential choice for pattern detection in wireless and mobile environments. Inference of behavior can, however, be simplified (S. Dattagupta and R. Sofia) and some activities can be performed locally, in mobile devices. Firstly, because it allows considering extraction of data in a more secure way, keeping user anonymity. Secondly, this approach reduces the cost associated with data transmission to a cloud. Therefore, the allocation of local computational tasks for the classification of sensed data is of key importance for mobile sensing systems.
• Supervised and unsupervised modeling. Still in this context we envision contributions in the context of supervised and unsupervised modelling techniques and statistical techniques (F. Costigliola, T. Almada, A. Fonseca, F. Duarte), e.g., for improving feedback (e.g. text analytics in sensing middleware, or a better adaptation of interfaces).

LONG-TERM PURPOSE:

A relevant aspect to address in the context of a next generation internet concerns data mining and analytics applied locally (e.g. in decentralized systems, or in cloudlets - rf. to the BEING project, ). In this context, eager classification models such as neural networks have been the preferential choice for pattern detection in wireless and mobile environments. Inference of behavior can, however, be simplified and some activities can be performed locally, in mobile devices. Firstly, because it allows to consider extraction of data in a more secure way, keeping user anonymity. Secondly, this approach reduces the cost associated with data transmission to a cloud. Therefore, the allocation of local computational tasks for the classification of sensed data is of key importance for mobile sensing systems.

Contributions in this area of confluence concern the assumption that contextual and behavior inference as well as predictive analytics shall most likely be residing on the fringes of the internet (cloudlets, mobile devices). Research on almost-reliable protocols, as well as on neural networks to assist linear programming is expected as follow-up of work already under development by several researchers in the unit. Still in this context we envision contributions in the context of supervised and unsupervised modeling techniques and statistical techniques, e.g., for improving feedback (e.g. text analytics in sensing middleware, or a better adaptation of interfaces).

Knowledge and Management Information Systems

2018-2019 Main Work Topics:

• Entrepreunerial culture (R. Ribeiro ). Contributions derived from this area of confluence concern models for a systematic analysis of ideas and relevancy in the context of business (e.g., to increase the success of pitching ideas to potential investors); alignment of information systems management with the business strategies; technology adoption modeling; models for technology transfer derived from COPELABS activities.

This area of confluence integrates the know-how of members that teach executive and middle-management courses in the context of LISS (Lusofona Information Systems School, ULHT). This confluence area shall therefore support COPELABS scientific activities in the context of transfer, entrepreneurial culture, IPR strategy as well as in assisting researchers to pursue on their own projects (pet projects) derived from technology created within the research context of COPELABS. Contributions derived from this area of confluence concern models for a systematic analysis of ideas and relevancy in the context of business (e.g., to increase the success of pitching ideas to potential investors); alignment of information systems management with the business strategies; technology adoption modeling; models for technology transfer derived from COPELABS activities.