Virtualizzazione di funzioni di rete per applicazioni Internet of Things (IoT) in città intelligenti.

Responsabili di progetto

Alberto Zanella, Milica Pejanovic-djurisic

Accordo

MONTENEGRO - MoS-not in force - Ministero della Scienza del Montenegro

Bando

CNR-MoS 2017-2018

Dipartimento

Ingegneria, ICT e tecnologie per l'energia e i trasporti

Area tematica

Ingegneria, ICT e tecnologie per l'energia e i trasporti

Stato del progetto

Nuovo

Proposta di ricerca

One of the challenges of 5G (5-th Generation) wireless networks will be the integration of mobile radio access with the Internet of Things (IoT) paradigm. Billions of objects, equipped with sensors and radio interfaces, will enrich our homes, public buildings and streets, with the aim of collecting and processing information from the environment, allowing a more efficient use of the resources (e.g., power, gas, water, etc.) and improving quality of life. In this smart city scenario, thousands of devices will exploit the network infrastructure for control and/or monitoring purposes.
Due to the huge amount of smart city application requirements, there is not a one-fits-all solution, but rather a plethora of vertical application-specific solutions, such as some standard solutions, like Zigbee and 6LowPAN, that have resulted in extremely fragmented context and market. Moreover, the very stringent requirements in terms of throughput and delay imposed by 5G, cannot be satisfied by many of the above mentioned solutions, due to the distributed nature of the protocols.

The above problems could be overcome by making IoT networks programmable and managing the control plane in a centralized way. The Software Defined Networking (SDN) paradigm and OpenFlow, the most popular instance of SDN, have been recently proposed to solve analogous issues in the wired domain. In OpenFlow the network nodes handle incoming packets as specified in the so-called Flow Table. Each entry of the Flow Table is related to a flow and contains a matching-rule, which specifies the values of the header fields that must be found in the packets belonging to the flow and the actions that must be executed on the packets of the flow (e.g., drop, forward to, etc.). If the Flow Table does not contain any entry specifying how to deal a certain packet, the node sends a request to a software entity called Controller that has a high-level abstraction of the network elements. The Controller replies with information required to fill a new Flow Table entry for handling the packet.
At the same time, network function virtualization (NFV) of IoT networks has also drawn significant attention
from both academia and industry. Virtualization technology abstracts the sensor resources as logical units and allows their efficient usage by multiple concurrent applications, even with conflicting requirements and goals. In this context, IoT devices, equipped with sensors and/or actuators, represent virtual network functions (VNF), made available to the final user.

This collaboration aims at designing and developing a prototype of an architecture for IoT virtualization, exploiting the presence of a Software-Defined Controller for IoT. In particular, we will consider a scenario where many IoT networks are deployed in different buildings or parts of a city and are composed of different devices providing various services. Service requests, coming from a final user, will be managed by a network function virtualization (NFV) orchestrator, in charge of i) finding the proper location of the service among those made available by the different IoT networks; ii) translating the request in a proper way to be processed by the IoT Controller; iii) interacting with the Controller to program the identified IoT network, such that the requested service could be delivered in agreement with the requested quality of service (QoS).

The latter approach will require:
1. Definition and implementation of a Virtualised Infrastructure Manager (VIM), being a component of the NFV orchestrator, in charge of receiving requests from the user, in the form of high-level intent-based description of a service (e.g., the level of luminosity of a given street, to be received within a given maximum delay), and of translating these requests in a suitable way to be processed by the Controller.
2. Definition and implementation of a proper strategy implemented at the Controller to: i) identify the location (in terms of IoT network to be triggered) of the requested service; ii) program the corresponding IoT network in order to guarantee the requested QoS.
3. Design and implementation of a programmable IoT network, where the control plane has been moved to the centralised Controller, and where routers just implement the data plane. This requires the Controller to have a good and updated vision of the full IoT network (e.g., list of neighbours of each node).

The above issues will be addressed in this collaboration and a prototype of the defined architecture will be developed. Moreover, a large experimental campaign will be performed, taking into account the specific smart city application scenario. In particular, two IoT networks, one in Bologna and one in Podgorica, will be deployed. Each of the two networks will be composed of tens of devices, connected to the same Coordinator and managed by a unique Controller, installed in a server. Devices will be located in the Universities buildings, generating the traffic identified for the specific smart city application to be addressed.

This project is part of a cooperation already established in the framework of the COST Action IRACON, where the EWG-IoT addresses topics perfectly in line with those of this project. In this framework, a preliminary architecture has been defined and the implementation of the Controller and of the programmable IoT network have been started. As far as the hardware is concerned, the research groups will take advantage of the outcomes of the NoE NEWCOM# project, funded by FP7, and the related EuWIn platform developed during the project and composed of 150 IEEE 802.15.4 devices.
As far as the quality of the consortium is concerned. The two research groups have a large experience in the field of wireless communications, wireless sensor networks, Internet of Things and software-defined networking. In particular, the main skills of CNR group are in the design and development of IoT networks for smart city applications, whi

Obiettivi della ricerca

This collaboration addresses a smart city scenario, where buildings are equipped with sensing devices, whose data are made available to a final user. The different IoT networks will be programmable and will be managed by a Controller, deciding about paths connecting devices, such as other PHY and MAC parameters (e.g., number of retransmission, transmit powers, etc.).
The main objectives are:
a) To review the already defined architecture in order to address smart city scenarios;
b) To review the strategy currently implemented at the IoT Controller, where a simple Dijkstra algorithm is running, in order to suitable program the IoT network addressed by the user;
c) To implement a prototype of the above architecture and components, taking advantage of the components already implemented in the framework of the IRACON collaboration;
d) To test the above architecture considering a smart city scenario.
The main outputs of this research projects will be research papers to be presented at International Conferences and Journals. Moreover, the outcome of the research will be also presented at the IRACON meetings.

Ultimo aggiornamento: 08/12/2024