Cognition-Based Networks: A New Perspective on Network Optimization Using Learning and Distributed Intelligence

Abstract

IEEE Access\ud Volume 3, 2015, Article number 7217798, Pages 1512-1530\ud Open Access\ud Cognition-based networks: A new perspective on network optimization using learning and distributed intelligence (Article)\ud Zorzi, M.a , Zanella, A.a, Testolin, A.b, De Filippo De Grazia, M.b, Zorzi, M.bc \ud a Department of Information Engineering, University of Padua, Padua, Italy \ud b Department of General Psychology, University of Padua, Padua, Italy \ud c IRCCS San Camillo Foundation, Venice-Lido, Italy \ud View additional affiliations\ud View references (107)\ud Abstract\ud In response to the new challenges in the design and operation of communication networks, and taking inspiration from how living beings deal with complexity and scalability, in this paper we introduce an innovative system concept called COgnition-BAsed NETworkS (COBANETS). The proposed approach develops around the systematic application of advanced machine learning techniques and, in particular, unsupervised deep learning and probabilistic generative models for system-wide learning, modeling, optimization, and data representation. Moreover, in COBANETS, we propose to combine this learning architecture with the emerging network virtualization paradigms, which make it possible to actuate automatic optimization and reconfiguration strategies at the system level, thus fully unleashing the potential of the learning approach. Compared with the past and current research efforts in this area, the technical approach outlined in this paper is deeply interdisciplinary and more comprehensive, calling for the synergic combination of expertise of computer scientists, communications and networking engineers, and cognitive scientists, with the ultimate aim of breaking new ground through a profound rethinking of how the modern understanding of cognition can be used in the management and optimization of telecommunication network

Applying Machine Learning Techniques to a Real Cognitive Network: File Transfer ETAs Prediction

“…For each time, we create a dataset varying the following network and topology parameters: transmit power {0 dBm, 5 dBm, 10 dBm, 15 dBm, 20 dBm}, WiFi transmission channel {1, 6, 11}, number of nodes simultaneously receiving data from the transmitter (from 1 to 4) and distance between the transmitter and the receivers {1 m, 5 m, 10 m}. Finally, for each configuration, we measure the transmission duration of a 100 MB file 3 .…”

“…Each configuration of the network has been tested 10 times, and the median of these results has been stored in the dataset, so as to reduce the variability effects of the transmission channels. Note that, although choosing the median of 10 experiments as an input for our algorithms could artificially reduce the level of randomness typical of wireless transmissions, we highlight that the ETA measurements belonging to the same set of experiments only have 2 USB version of ath9k 3 The average size of a music video in full HD (1080p) resolution low standard deviation: a separate paper taking into account all these measurements is currently in preparation. The dataset is thus composed by M = 540 examples.…”

“…Cognitive networking (CN) [1]- [3] is a network paradigm that deals with how wireless systems learn the relationships between the network status and behavior on one side, and its performance on the other, and plan and make decisions to achieve local, end-to-end, and network-wide goals. Differently from cognitive radio (CR) networking, CN tries to jointly optimize the different layers of the protocol stack.…”

Softwarization and Optimization for Sustainable Future Mobile Networks: A Survey

“…Such learning platforms should account for noisy nature of data collection and ensure prediction of what measurements are to be expected. Such network wide M L based cognition platform is proposed in [86]. The authors identified how generative deep neural networks (GDNN) are suited for network wide cognition and full self-organization capability.…”

Sub-GHz LPWAN Network Coexistence, Management and Virtualization: An Overview and Open Research Challenges

“…The neural network is trained to identify how environmental measurements and the status of the network affect the performance experienced on different channels. In [52] [54], where a locating array is used to identify the 5 most influential parameters out of an initial design space of 24 parameters. As such, by applying dimensionality reduction techniques before model creation, the exploration time can significantly be reduced.…”