Skip to main content

Erol Gelenbe

Imperial College London, Electrical and Electronic Engineering, Faculty Member

Followers

439

Following

115

Co-authors

32

Public Views

Erol Gelenbe is a Chaired Professor (Dennis Gabor Chair) at Imperial College, London. Known for his research and leadership in the field of computers and communication networks, and for inventing mathematical models such as G-Networks and the Random Neural Network, and inventing computer architectures for Voice-over-Packets, the Cognitive Packet Network routing protocol, the SYCOMORE voice-packet switch and the FLEXSIM object oriented discrete event simulator, he has graduated 77 PhDs. He received three honorary doctorates and several prestigious prizes and is an elected Fellow of the ACM (USA, Fellow of IEEE (USA), the National Academy of Technologies (Engineering) of France, and of the National Science Academies of Belgium, Hungary, Poland and Turkey.
Supervisors: Edward J. Smith Jr., Polytechnic Institute of New York University and Jacques-Louis Lions, College de France and University of Paris VI (Pierre et Marie Curie)

less

Swiss Federal Institute of Technology (ETH)

Andrey Gavrilov

Novosibirsk State Technical University

Computer Science & Information Technology (CS & IT) Computer Science Conference Proceedings (CSCP)

Michele D'Asaro

Radboud University Nijmegen

Kenneth Kreutz-delgado

University of California, San Diego

Gert Cauwenberghs

University of California, San Diego

Dalian University of Technology

Ramesh Chandra Agrawal

Interests

Uploads

Papers by Erol Gelenbe

Accurate and Energy-Efficient Classification with Spiking Random Neural Network: Corrected and Expanded Version

Probability in the Engineering and Informational Sciences, 2019

Artificial Neural Network (ANN) based techniques have dominated state-of-the-art results in most ... more Artificial Neural Network (ANN) based techniques have dominated state-of-the-art results in most problems related to computer vision, audio recognition, and natural language processing in the past few years, resulting in strong industrial adoption from all leading technology companies worldwide. One of the major obstacles that have historically delayed large scale adoption of ANNs is the huge computational and power costs associated with training and testing (deploying) them. In the meantime , Neuromorphic Computing platforms have recently achieved remarkable performance running more bio-realistic Spiking Neural Networks at high throughput and very low power consumption making them a natural alternative to ANNs. Here, we propose using the Random Neural Network (RNN), a spiking neural network with both theoretical and practical appealing properties, as a general purpose classifier that can match the classification power of ANNs on a number of tasks while enjoying all the features of a spiking neural network. This is demonstrated on a number of real-world classification datasets.

electronics A Survey of Algorithms and Systems for Evacuating People in Confined Spaces

The frequency, destruction and costs of natural and human-made disasters in modern highly-populat... more The frequency, destruction and costs of natural and human-made disasters in modern highly-populated societies have resulted in research on emergency evacuation and wayfinding, which has drawn considerable attention. The subject is now a multidisciplinary area of research where information and communication technologies (ICT), and in particular the Internet of Things (IoT), have a significant impact on sensing and computing dynamic reactions that mitigate or prevent the worst outcomes of disasters. This paper offers state-of-the-art knowledge in this area so as to share ongoing research results, identify the research gaps and address the need for future research. We present a comprehensive review of research on emergency evacuation and wayfinding, focusing on the algorithmic and system design aspects. Starting from the history of emergency management research, we identify the emerging challenges concerning system optimisation, evacuee behaviour optimisation and data analysis, and the additional energy consumption by ICT equipment that operates the emergency management infrastructure.

Research and Innovation Action for the Security of the Internet of Things: The SerIoT Project

The Internet of Things (IoT) was born in the mid 2010's, when the threshold of connecting more ob... more The Internet of Things (IoT) was born in the mid 2010's, when the threshold of connecting more objects than people to the Inter-net, was crossed. Thus, attacks and threats on the content and quality of service of the IoT platforms can have economic, energetic and physical security consequences that go way beyond the traditional Internet's lack of security, and way beyond the threats posed by attacks to mobile tele-phony. Thus, this paper describes the H2020 project "Secure and Safe Internet of Things" (SerIoT) which will optimize the information security in IoT platforms and networks in a holistic, cross-layered manner (i.e. IoT platforms and devices, honeypots, SDN routers and operator's controller) in order to offer a secure SerIoT platform that can be used to implement secure IoT platforms and networks anywhere and everywhere.

Top-k Query Result Completeness Verification in Tiered Sensor Networks

Storage nodes are expected to be placed as an intermediate tier of large scale sensor networks fo... more Storage nodes are expected to be placed as an intermediate tier of large scale sensor networks for caching the collected sensor readings and responding to queries with benefits of power and storage saving for ordinary sensors. Nevertheless, an important issue is that the compromised storage node may not only cause the privacy problem, but also return fake/incomplete query results. We propose a simple yet effective dummy reading-based anonymization framework, under which the query result integrity can be guaranteed by our proposed verifiable top-k query (VQ) schemes. Compared with existing works, the VQ schemes have a fundamentally different design philosophy and achieve the lower communication complexity at the cost of slight detection capability degradation. Analytical studies, numerical simulations, and prototype implementations are conducted to demonstrate the practicality of our proposed methods.

ScienceDirect First workshop on Secure and Efficient Deployment of IoT (SEDIT 2018) Deep Learning with Dense Random Neural Network for Detecting Attacks against IoT-connected Home Environments

In this paper, we analyze the network attacks that can be launched against Internet of Things (Io... more In this paper, we analyze the network attacks that can be launched against Internet of Things (IoT) gateways, identify the relevant metrics to detect them, and explain how they can be computed from packet captures. We then present the principles and design of a deep learning-based approach using dense random neural networks (RNN) for the online detection of network attacks. Empirical validation results on packet captures in which attacks are inserted show that the Dense RNN correctly detects attacks. However our experiments show that a simple threshold detector also provides results of comparable accuracy on the same data set.

Security for Internet of Things: The SerIoT Project 1 st Erol Gelenbe

by Erol Gelenbe and J. Domanska

Accurate, Energy-Efficient Classification with Spiking Random Neural Network

Artificial Neural Networks (ANNs) based techniques have dominated state-of-the-art 1 results in m... more Artificial Neural Networks (ANNs) based techniques have dominated state-of-the-art 1 results in most problems related to computer vision, audio recognition, and natural language 2 processing in the past few years, resulting in strong industrial adoption from all leading technology 3 companies worldwide. One of the major obstacles that have historically delayed large scale adoption 4 of ANNs is the huge computational and power costs associated with training and testing (deploying) 5 them. In the meantime , Neuromorphic Computing platforms have recently achieved remarkable 6 performance running the bio-realistic Spiking Neural Networks at high throughput and very low 7 power consumption making them a natural alternative to ANNs if they could match their classification 8 performance. Here, we propose using the Random Neural Network (RNN), a spiking neural network 9 with both theoretical and practical appealing properties, as a general purpose classifier that can match 10 the classification power of ANNs on a number of tasks while enjoying all the features of being a 11 spiking neural network. This is demonstrated on a number of real-world classification datasets. 12

Static Analysis-Based Approaches for Secure Software Development

Software security is a matter of major concern for software development enterprises that wish to ... more Software security is a matter of major concern for software development enterprises that wish to deliver highly secure software products to their customers. Static analysis is considered one of the most effective mechanisms for adding security to software products. The multitude of static analysis tools that are available provide a large number of raw results that may contain security-relevant information, which may be useful for the production of secure software. Several mechanisms that can facilitate the production of both secure and reliable software applications have been proposed over the years. In this paper, two such mechanisms, particularly the vulnerability prediction models (VPMs) and the optimum checkpoint recommendation (OCR) mechanisms, are theoretically examined, while their potential improvement by using static analysis is also investigated. In particular, we review the most significant contributions regarding these mechanisms, identify their most important open issues, and propose directions for future research, emphasizing on the potential adoption of static analysis for addressing the identified open issues. Hence, this paper can act as a reference for researchers that wish to contribute in these subfields, in order to gain solid understanding of the existing solutions and their open issues that require further research.

Peer Prediction Based Trustworthiness Evaluation and Trustworthy Service Rating in Social Networks

by Erol Gelenbe and Vince Poor

With the development of online applications based on the social network, many different approache... more With the development of online applications based on the social network, many different approaches of service to achieve these applications have emerged. Users' reporting and sharing of their consumption experience or opinion can be utilized to rate the quality of different approaches of online services. How to ensure the authenticity of the users' reports and identify malicious ones with cheating reports become important issues to achieve an accurate service rating. In this paper, we provide a private-prior peer prediction mechanism based trustworthy service rating system with a data processing center (DPC), which requires users to report to it with their prior and posterior believes that their peer users will report a high quality opinion of the service. The DPC evaluates users' trustworthiness with their reports by applying the strictly proper scoring rule, and removes reports received from users with low trustworthiness from the service rating procedure. This peer prediction method is incentive compatible and able to motivate users to report honestly. In addition, to identify malicious users and bad-functioning/unreliable users with high error rate of quality judgement, an unreliability index is proposed in this paper to evaluate the uncertainty of reports. Reports with high unreliability values will also be excluded from the service rating system. By combining the trustworthiness and unreliability, malicious users will face a dilemma that they cannot receive a high trustworthiness and low unreliability at the same time when they report falsely. Simulation results indicate that the proposed peer prediction based trustworthy service rating can identify malicious and unreliable behaviours effectively, and motivate users to report truthfully. The relatively high service rating accuracy can be achieved by the proposed system.

SerIoT: An H2020 Research and Innovation Action to Assure the Security of the Internet of Things

The Internet of Things (IoT) was born in the mid 2010's, when the threshold of connecting more ob... more The Internet of Things (IoT) was born in the mid 2010's, when the threshold of connecting more objects than people to the Internet, was crossed. Thus, attacks and threats on the content and quality of service of the IoT platforms can have economic, energetic and physical security consequences that go way beyond the traditional Internet's lack of security, and way beyond the threats posed by attacks to mobile telephony. Thus, this paper describes the H2020 project "Secure and Safe Internet of Things"(SerIoT) which will optimize the information security in IoT platforms and networks in a holistic, cross-layered manner (i.e. IoT platforms and devices, honeypots, SDN routers and operator's controller) in order to offer a secure SerIoT platform that can be used to implement secure IoT platforms and networks anywhere and everywhere.

European Cybersecurity Research and the SerIoT Project

This paper briefly reviews some recent research in Cybersecurity in Europe funded by the European... more This paper briefly reviews some recent research in Cybersecurity in Europe funded by the European Commission in areas such as mobile telephony, networked health systems, the Internet of Things. We then outline the objectives of the SerIoT Project which started in 2018 to address the security needs of fields such as Smart Cities, Smart Transportation Systems, Supply Chains and Industrial Informatics.

SerIoT -Secure and Safe Internet of Things View project Cognitive Packet Network View project Data Transaction Modeling in Mobile Networks: Contract Mechanism and Performance Analaysis

We consider auction mechanism design and performance analysis for data transactions in mobile soc... more We consider auction mechanism design and performance analysis for data transactions in mobile social networks. Existing mobile network plans can result in some users ending a monthly plan with excess data, while others may have to pay a costly fee to buy more data. Thus we suggest data auctions with a single seller, or a multiple-seller networked data auction, that operate in mobile social networks, to deal with the asymmetry between extra unused data resources and urgent data demands. Based on earlier work on the analysis of auctions, we design the data transaction mechanism, and summarise the analysis on state transmission, stationary probabilities of the system, and the expected income for data sellers. To improve the efficiency and performance of the system, a socially-aware mobility model is also proposed. The proposed data auction mechanisms and friendship-based mobility model are then simulated as operating on Flickr, a real-world online social network database. Results show that the number of data bidders in different auctions can be balanced through the proposed mobility model, and also increase the income per unit time of sellers in the networked data auction.

Energy Loss through Standby and Leakage in Energy Harvesting Wireless Sensors

We present a performance model for an energy harvesting wireless sensor node in which data gather... more We present a performance model for an energy harvesting wireless sensor node in which data gathering and harvesting are slow random processes as compared to fast wireless communications. We assume that the system will use stored energy when collecting data in standby, and that energy will leak from capacitors and batteries. In the presence of these imperfections we derive the system's packet transmission capacity when its packet storage buffer and its energy storage unit have a finite capacity that may lead to both data packet overflows, and the loss of incoming energy in addition to standby losses. We also consider an infinite capacity model which operates in the presence of transmission errors due to channel noise and interference.

The Deep Learning Random Neural Network with a Management Cluster

We propose the Random Neural Network with Deep Learning Clusters that emulates the way the brain ... more We propose the Random Neural Network with Deep Learning Clusters that emulates the way the brain takes decisions. We apply our model to measure and evaluate Web result relevance by associating each Deep Learning Cluster with a different Web Search Engine; in addition, we include a Deep Learning Cluster to perform as a Management Cluster that decides the final result relevance based on the inputs from the Deep Learning clusters. Our algorithm reorders the Web results obtained from different Web Search Engines after a user introduces a query. We evaluate the performance of the Management Cluster when included as an additional layer to the Deep Learning Clusters. On average; our Management Cluster improves result relevance; its inclusion increments the overall result relevance quality.

Demonstrating Voice over an Autonomic Network

We demonstrate experimentally how an Autonomic Network based on the CPN protocol can provide the ... more We demonstrate experimentally how an Autonomic Network based on the CPN protocol can provide the Quality of Service (QoS) required by voice communications. The implementation uses Reinforcement Learning to dynamically seek paths that meet the quality requirements of voice communications. Measurements of packet delay, jitter, and loss illustrate the performance obtained from the system.

Energy and QoS in a cognitive channel Energy and QoS in a Cognitive Channel

Smart network users can use sensing and intelligence to optimise various parameters for their com... more Smart network users can use sensing and intelligence to optimise various parameters for their communications as in CSMA channels. Though this paper fo-cuses on energy consumption per successfully transmitted packet as the primary QoS metric, we also examine how the different QoS metrics, such as throughput and delay, also interact with energy consumption, and show how trade-offs can be effected among them. The approach combines a queueing theoretic analysis with the sensing and error control effects in the channel in the presence of interfering communications.

G-Networks Batch Removal

Communication Protocol Coupled Queues

Minimum Graph Covering

Steady-state solution of probabilistic gene regulatory networks

We introduce a probability model for gene regulatory networks, based on a system of Chapman-Kolmo... more We introduce a probability model for gene regulatory networks, based on a system of Chapman-Kolmogorov equations that represent the dynamics of the concentration levels of each agent in the network. This unifying approach includes the representation of excitatory and inhibitory interactions between agents, second-order interactions which allow any two agents to jointly act on other agents, and Boolean dependencies between agents. The probability model represents the concentration or quantity of each agent, and we obtain the equilibrium solution for the joint probability distribution of each of the concentrations. The result is an exact solution in "product form," where the joint equilibrium probability distribution of the concentration for each gene is the product of the marginal distribution for each of the concentrations. The analysis we present yields the probability distribution of the concentration or quantity of all of the agents in a network that includes both logical dependencies and excitatory-inhibitory relationships between agents. I. GENE REGULATORY NETWORKS Kaufman 1 pioneered models of gene regulatory networks 2 that have been extended 3,4 in order to include logical dependencies between agents 5-8, as well as sto-chastic dynamics 9-11. In this paper we develop a unifying approach to model the noisy behavior of regulatory networks that includes i exci-tatory and ii inhibitory interactions between agents, and iii second-order interactions which allow any two agents to jointly act on other agents. We also show that Boolean dependencies between agents can be modeled with our approach by using second-order interactions. The model studied in this paper represents the concentration levels or quantity of each agent in the network. All transitions in the model are probabilistic. Time is represented via random transition times whose average value depends on which agents are involved in each transition. The work in 12,13 is a precursor of the approach that we develop here. The present paper extends our prior work on G networks 14,15, so as to compute the probability of activation of the agents in the presence of complex interactions. The model in this paper differs from probabilistic Boolean networks 10 in that we propose an integer valued concentration level for each agent i, denoted by K i t, and we study the stochastic dynamical behavior of the vector Kt whose elements are the K i t, with the probability distribution Pk , t = PKt = k. On the other hand, we define a mapping of the variables K i t into binary variables B i t such that B i t =1 if K i t 0, and B i t =0 if K i t = 0, and also compute the steady-state probabilities of the B i t from the corresponding distribution for the Pk , t. However, we do not deal directly with the dynamics of the vector Bt whose elements are the B i t. In our model Kt is a Markov chain in continuous time, but Bt is not a Markov chain. In 16 a deterministic population model is considered; it uses nonlinear ordinary differential equations Eq. 1 in 16 to represent the concentration or quantity of different genes. The approach in 16 is similar to the use of the general mass equations GMA of chemistry, and variability due to biology and measurement noise is represented in 16 by modifying the parameter values in the data sets. Our approach uses a probabilistic model similar to the chemical master equations CME of chemistry, and noise is intrinsi-cally part of the model. In chemistry, the GMA are a "mac-roscopic" deterministic approximation of the "microscopic" probabilistic representation provided by the CME. The work of Ribeiro et al. 11 also considers the latter approach, and in 11 the system is analyzed using Monte Carlo simulations , while our work pursues an analytic approach. Also, 11 describes the logical dependencies between agents via rate equations, while here we present both the probabilistic CME and derive explicit Boolean dependencies between these equations. Since practical measurements with microarrays will deal with large populations of cells each of whose individual instantaneous behavior may not be synchronized, in 17 the effect of the variation in the number of cells which have a given gene expression at a given measurement instant is studied; signal processing techniques are used to derive the correct gene expression for cyclic gene expression from a large number of cells with specific reference to a single gene. This paper presents a model based on a single cell and multiple agents, and includes the time-dependent probabilistic dynamics as presented in the differential equation 5. II. REGULATORY NETWORKS AND G NETWORKS We will first begin by presenting the model that we propose , which is based on G networks 14,18,15,19, which are probabilistic dynamical models with an unbounded discrete state space, operating in continuous time. The model is composed of the following: i Agents, which are the primary objects of interest; they represent genes or other active biochemical or living objects whose levels of activity we wish to represent. ii Gates, which represent the interactions between *[email protected]

Accurate and Energy-Efficient Classification with Spiking Random Neural Network: Corrected and Expanded Version

Probability in the Engineering and Informational Sciences, 2019

Artificial Neural Network (ANN) based techniques have dominated state-of-the-art results in most ... more Artificial Neural Network (ANN) based techniques have dominated state-of-the-art results in most problems related to computer vision, audio recognition, and natural language processing in the past few years, resulting in strong industrial adoption from all leading technology companies worldwide. One of the major obstacles that have historically delayed large scale adoption of ANNs is the huge computational and power costs associated with training and testing (deploying) them. In the meantime , Neuromorphic Computing platforms have recently achieved remarkable performance running more bio-realistic Spiking Neural Networks at high throughput and very low power consumption making them a natural alternative to ANNs. Here, we propose using the Random Neural Network (RNN), a spiking neural network with both theoretical and practical appealing properties, as a general purpose classifier that can match the classification power of ANNs on a number of tasks while enjoying all the features of a spiking neural network. This is demonstrated on a number of real-world classification datasets.

electronics A Survey of Algorithms and Systems for Evacuating People in Confined Spaces

The frequency, destruction and costs of natural and human-made disasters in modern highly-populat... more The frequency, destruction and costs of natural and human-made disasters in modern highly-populated societies have resulted in research on emergency evacuation and wayfinding, which has drawn considerable attention. The subject is now a multidisciplinary area of research where information and communication technologies (ICT), and in particular the Internet of Things (IoT), have a significant impact on sensing and computing dynamic reactions that mitigate or prevent the worst outcomes of disasters. This paper offers state-of-the-art knowledge in this area so as to share ongoing research results, identify the research gaps and address the need for future research. We present a comprehensive review of research on emergency evacuation and wayfinding, focusing on the algorithmic and system design aspects. Starting from the history of emergency management research, we identify the emerging challenges concerning system optimisation, evacuee behaviour optimisation and data analysis, and the additional energy consumption by ICT equipment that operates the emergency management infrastructure.

Research and Innovation Action for the Security of the Internet of Things: The SerIoT Project

The Internet of Things (IoT) was born in the mid 2010's, when the threshold of connecting more ob... more The Internet of Things (IoT) was born in the mid 2010's, when the threshold of connecting more objects than people to the Inter-net, was crossed. Thus, attacks and threats on the content and quality of service of the IoT platforms can have economic, energetic and physical security consequences that go way beyond the traditional Internet's lack of security, and way beyond the threats posed by attacks to mobile tele-phony. Thus, this paper describes the H2020 project "Secure and Safe Internet of Things" (SerIoT) which will optimize the information security in IoT platforms and networks in a holistic, cross-layered manner (i.e. IoT platforms and devices, honeypots, SDN routers and operator's controller) in order to offer a secure SerIoT platform that can be used to implement secure IoT platforms and networks anywhere and everywhere.

Top-k Query Result Completeness Verification in Tiered Sensor Networks

Storage nodes are expected to be placed as an intermediate tier of large scale sensor networks fo... more Storage nodes are expected to be placed as an intermediate tier of large scale sensor networks for caching the collected sensor readings and responding to queries with benefits of power and storage saving for ordinary sensors. Nevertheless, an important issue is that the compromised storage node may not only cause the privacy problem, but also return fake/incomplete query results. We propose a simple yet effective dummy reading-based anonymization framework, under which the query result integrity can be guaranteed by our proposed verifiable top-k query (VQ) schemes. Compared with existing works, the VQ schemes have a fundamentally different design philosophy and achieve the lower communication complexity at the cost of slight detection capability degradation. Analytical studies, numerical simulations, and prototype implementations are conducted to demonstrate the practicality of our proposed methods.

ScienceDirect First workshop on Secure and Efficient Deployment of IoT (SEDIT 2018) Deep Learning with Dense Random Neural Network for Detecting Attacks against IoT-connected Home Environments

In this paper, we analyze the network attacks that can be launched against Internet of Things (Io... more In this paper, we analyze the network attacks that can be launched against Internet of Things (IoT) gateways, identify the relevant metrics to detect them, and explain how they can be computed from packet captures. We then present the principles and design of a deep learning-based approach using dense random neural networks (RNN) for the online detection of network attacks. Empirical validation results on packet captures in which attacks are inserted show that the Dense RNN correctly detects attacks. However our experiments show that a simple threshold detector also provides results of comparable accuracy on the same data set.

Security for Internet of Things: The SerIoT Project 1 st Erol Gelenbe

by Erol Gelenbe and J. Domanska

Accurate, Energy-Efficient Classification with Spiking Random Neural Network

Artificial Neural Networks (ANNs) based techniques have dominated state-of-the-art 1 results in m... more Artificial Neural Networks (ANNs) based techniques have dominated state-of-the-art 1 results in most problems related to computer vision, audio recognition, and natural language 2 processing in the past few years, resulting in strong industrial adoption from all leading technology 3 companies worldwide. One of the major obstacles that have historically delayed large scale adoption 4 of ANNs is the huge computational and power costs associated with training and testing (deploying) 5 them. In the meantime , Neuromorphic Computing platforms have recently achieved remarkable 6 performance running the bio-realistic Spiking Neural Networks at high throughput and very low 7 power consumption making them a natural alternative to ANNs if they could match their classification 8 performance. Here, we propose using the Random Neural Network (RNN), a spiking neural network 9 with both theoretical and practical appealing properties, as a general purpose classifier that can match 10 the classification power of ANNs on a number of tasks while enjoying all the features of being a 11 spiking neural network. This is demonstrated on a number of real-world classification datasets. 12

Static Analysis-Based Approaches for Secure Software Development

Software security is a matter of major concern for software development enterprises that wish to ... more Software security is a matter of major concern for software development enterprises that wish to deliver highly secure software products to their customers. Static analysis is considered one of the most effective mechanisms for adding security to software products. The multitude of static analysis tools that are available provide a large number of raw results that may contain security-relevant information, which may be useful for the production of secure software. Several mechanisms that can facilitate the production of both secure and reliable software applications have been proposed over the years. In this paper, two such mechanisms, particularly the vulnerability prediction models (VPMs) and the optimum checkpoint recommendation (OCR) mechanisms, are theoretically examined, while their potential improvement by using static analysis is also investigated. In particular, we review the most significant contributions regarding these mechanisms, identify their most important open issues, and propose directions for future research, emphasizing on the potential adoption of static analysis for addressing the identified open issues. Hence, this paper can act as a reference for researchers that wish to contribute in these subfields, in order to gain solid understanding of the existing solutions and their open issues that require further research.

Peer Prediction Based Trustworthiness Evaluation and Trustworthy Service Rating in Social Networks

by Erol Gelenbe and Vince Poor

With the development of online applications based on the social network, many different approache... more With the development of online applications based on the social network, many different approaches of service to achieve these applications have emerged. Users' reporting and sharing of their consumption experience or opinion can be utilized to rate the quality of different approaches of online services. How to ensure the authenticity of the users' reports and identify malicious ones with cheating reports become important issues to achieve an accurate service rating. In this paper, we provide a private-prior peer prediction mechanism based trustworthy service rating system with a data processing center (DPC), which requires users to report to it with their prior and posterior believes that their peer users will report a high quality opinion of the service. The DPC evaluates users' trustworthiness with their reports by applying the strictly proper scoring rule, and removes reports received from users with low trustworthiness from the service rating procedure. This peer prediction method is incentive compatible and able to motivate users to report honestly. In addition, to identify malicious users and bad-functioning/unreliable users with high error rate of quality judgement, an unreliability index is proposed in this paper to evaluate the uncertainty of reports. Reports with high unreliability values will also be excluded from the service rating system. By combining the trustworthiness and unreliability, malicious users will face a dilemma that they cannot receive a high trustworthiness and low unreliability at the same time when they report falsely. Simulation results indicate that the proposed peer prediction based trustworthy service rating can identify malicious and unreliable behaviours effectively, and motivate users to report truthfully. The relatively high service rating accuracy can be achieved by the proposed system.

SerIoT: An H2020 Research and Innovation Action to Assure the Security of the Internet of Things

The Internet of Things (IoT) was born in the mid 2010's, when the threshold of connecting more ob... more The Internet of Things (IoT) was born in the mid 2010's, when the threshold of connecting more objects than people to the Internet, was crossed. Thus, attacks and threats on the content and quality of service of the IoT platforms can have economic, energetic and physical security consequences that go way beyond the traditional Internet's lack of security, and way beyond the threats posed by attacks to mobile telephony. Thus, this paper describes the H2020 project "Secure and Safe Internet of Things"(SerIoT) which will optimize the information security in IoT platforms and networks in a holistic, cross-layered manner (i.e. IoT platforms and devices, honeypots, SDN routers and operator's controller) in order to offer a secure SerIoT platform that can be used to implement secure IoT platforms and networks anywhere and everywhere.

European Cybersecurity Research and the SerIoT Project

This paper briefly reviews some recent research in Cybersecurity in Europe funded by the European... more This paper briefly reviews some recent research in Cybersecurity in Europe funded by the European Commission in areas such as mobile telephony, networked health systems, the Internet of Things. We then outline the objectives of the SerIoT Project which started in 2018 to address the security needs of fields such as Smart Cities, Smart Transportation Systems, Supply Chains and Industrial Informatics.

SerIoT -Secure and Safe Internet of Things View project Cognitive Packet Network View project Data Transaction Modeling in Mobile Networks: Contract Mechanism and Performance Analaysis

We consider auction mechanism design and performance analysis for data transactions in mobile soc... more We consider auction mechanism design and performance analysis for data transactions in mobile social networks. Existing mobile network plans can result in some users ending a monthly plan with excess data, while others may have to pay a costly fee to buy more data. Thus we suggest data auctions with a single seller, or a multiple-seller networked data auction, that operate in mobile social networks, to deal with the asymmetry between extra unused data resources and urgent data demands. Based on earlier work on the analysis of auctions, we design the data transaction mechanism, and summarise the analysis on state transmission, stationary probabilities of the system, and the expected income for data sellers. To improve the efficiency and performance of the system, a socially-aware mobility model is also proposed. The proposed data auction mechanisms and friendship-based mobility model are then simulated as operating on Flickr, a real-world online social network database. Results show that the number of data bidders in different auctions can be balanced through the proposed mobility model, and also increase the income per unit time of sellers in the networked data auction.

Energy Loss through Standby and Leakage in Energy Harvesting Wireless Sensors

We present a performance model for an energy harvesting wireless sensor node in which data gather... more We present a performance model for an energy harvesting wireless sensor node in which data gathering and harvesting are slow random processes as compared to fast wireless communications. We assume that the system will use stored energy when collecting data in standby, and that energy will leak from capacitors and batteries. In the presence of these imperfections we derive the system's packet transmission capacity when its packet storage buffer and its energy storage unit have a finite capacity that may lead to both data packet overflows, and the loss of incoming energy in addition to standby losses. We also consider an infinite capacity model which operates in the presence of transmission errors due to channel noise and interference.

The Deep Learning Random Neural Network with a Management Cluster

We propose the Random Neural Network with Deep Learning Clusters that emulates the way the brain ... more We propose the Random Neural Network with Deep Learning Clusters that emulates the way the brain takes decisions. We apply our model to measure and evaluate Web result relevance by associating each Deep Learning Cluster with a different Web Search Engine; in addition, we include a Deep Learning Cluster to perform as a Management Cluster that decides the final result relevance based on the inputs from the Deep Learning clusters. Our algorithm reorders the Web results obtained from different Web Search Engines after a user introduces a query. We evaluate the performance of the Management Cluster when included as an additional layer to the Deep Learning Clusters. On average; our Management Cluster improves result relevance; its inclusion increments the overall result relevance quality.

Demonstrating Voice over an Autonomic Network

We demonstrate experimentally how an Autonomic Network based on the CPN protocol can provide the ... more We demonstrate experimentally how an Autonomic Network based on the CPN protocol can provide the Quality of Service (QoS) required by voice communications. The implementation uses Reinforcement Learning to dynamically seek paths that meet the quality requirements of voice communications. Measurements of packet delay, jitter, and loss illustrate the performance obtained from the system.

Energy and QoS in a cognitive channel Energy and QoS in a Cognitive Channel

Smart network users can use sensing and intelligence to optimise various parameters for their com... more Smart network users can use sensing and intelligence to optimise various parameters for their communications as in CSMA channels. Though this paper fo-cuses on energy consumption per successfully transmitted packet as the primary QoS metric, we also examine how the different QoS metrics, such as throughput and delay, also interact with energy consumption, and show how trade-offs can be effected among them. The approach combines a queueing theoretic analysis with the sensing and error control effects in the channel in the presence of interfering communications.

G-Networks Batch Removal

Communication Protocol Coupled Queues

Minimum Graph Covering

Steady-state solution of probabilistic gene regulatory networks

We introduce a probability model for gene regulatory networks, based on a system of Chapman-Kolmo... more We introduce a probability model for gene regulatory networks, based on a system of Chapman-Kolmogorov equations that represent the dynamics of the concentration levels of each agent in the network. This unifying approach includes the representation of excitatory and inhibitory interactions between agents, second-order interactions which allow any two agents to jointly act on other agents, and Boolean dependencies between agents. The probability model represents the concentration or quantity of each agent, and we obtain the equilibrium solution for the joint probability distribution of each of the concentrations. The result is an exact solution in "product form," where the joint equilibrium probability distribution of the concentration for each gene is the product of the marginal distribution for each of the concentrations. The analysis we present yields the probability distribution of the concentration or quantity of all of the agents in a network that includes both logical dependencies and excitatory-inhibitory relationships between agents. I. GENE REGULATORY NETWORKS Kaufman 1 pioneered models of gene regulatory networks 2 that have been extended 3,4 in order to include logical dependencies between agents 5-8, as well as sto-chastic dynamics 9-11. In this paper we develop a unifying approach to model the noisy behavior of regulatory networks that includes i exci-tatory and ii inhibitory interactions between agents, and iii second-order interactions which allow any two agents to jointly act on other agents. We also show that Boolean dependencies between agents can be modeled with our approach by using second-order interactions. The model studied in this paper represents the concentration levels or quantity of each agent in the network. All transitions in the model are probabilistic. Time is represented via random transition times whose average value depends on which agents are involved in each transition. The work in 12,13 is a precursor of the approach that we develop here. The present paper extends our prior work on G networks 14,15, so as to compute the probability of activation of the agents in the presence of complex interactions. The model in this paper differs from probabilistic Boolean networks 10 in that we propose an integer valued concentration level for each agent i, denoted by K i t, and we study the stochastic dynamical behavior of the vector Kt whose elements are the K i t, with the probability distribution Pk , t = PKt = k. On the other hand, we define a mapping of the variables K i t into binary variables B i t such that B i t =1 if K i t 0, and B i t =0 if K i t = 0, and also compute the steady-state probabilities of the B i t from the corresponding distribution for the Pk , t. However, we do not deal directly with the dynamics of the vector Bt whose elements are the B i t. In our model Kt is a Markov chain in continuous time, but Bt is not a Markov chain. In 16 a deterministic population model is considered; it uses nonlinear ordinary differential equations Eq. 1 in 16 to represent the concentration or quantity of different genes. The approach in 16 is similar to the use of the general mass equations GMA of chemistry, and variability due to biology and measurement noise is represented in 16 by modifying the parameter values in the data sets. Our approach uses a probabilistic model similar to the chemical master equations CME of chemistry, and noise is intrinsi-cally part of the model. In chemistry, the GMA are a "mac-roscopic" deterministic approximation of the "microscopic" probabilistic representation provided by the CME. The work of Ribeiro et al. 11 also considers the latter approach, and in 11 the system is analyzed using Monte Carlo simulations , while our work pursues an analytic approach. Also, 11 describes the logical dependencies between agents via rate equations, while here we present both the probabilistic CME and derive explicit Boolean dependencies between these equations. Since practical measurements with microarrays will deal with large populations of cells each of whose individual instantaneous behavior may not be synchronized, in 17 the effect of the variation in the number of cells which have a given gene expression at a given measurement instant is studied; signal processing techniques are used to derive the correct gene expression for cyclic gene expression from a large number of cells with specific reference to a single gene. This paper presents a model based on a single cell and multiple agents, and includes the time-dependent probabilistic dynamics as presented in the differential equation 5. II. REGULATORY NETWORKS AND G NETWORKS We will first begin by presenting the model that we propose , which is based on G networks 14,18,15,19, which are probabilistic dynamical models with an unbounded discrete state space, operating in continuous time. The model is composed of the following: i Agents, which are the primary objects of interest; they represent genes or other active biochemical or living objects whose levels of activity we wish to represent. ii Gates, which represent the interactions between *[email protected]

Accurate and Energy-Efficient Classification with Spiking Random Neural Network: Corrected and Expanded Version

by Erol Gelenbe and Mohamed Youssef Bassyouni

arXiv:1906.08864, 2019

Artificial Neural Network (ANN) based techniques have dominated state-of-the-art results in most ... more Artificial Neural Network (ANN) based techniques have dominated state-of-the-art results in most problems related to computer vision, audio recognition, and natural language processing in the past few years, resulting in strong industrial adoption from all leading technology companies worldwide. One of the major obstacles that have historically delayed large scale adoption of ANNs is the huge computational and power costs associated with training and testing (deploying) them. In the meantime , Neuromorphic Computing platforms have recently achieved remarkable performance running more bio-realistic Spiking Neural Networks at high throughput and very low power consumption making them a natural alternative to ANNs. Here, we propose using the Random Neural Network (RNN), a spiking neural network with both theoretical and practical appealing properties, as a general purpose classifier that can match the classification power of ANNs on a number of tasks while enjoying all the features of a spiking neural network. This is demonstrated on a number of real-world classification datasets.

Modelling the Impact of Cyber-Attacks on Web Based Sales

Unpublished, 2019

Performance Modelling to predict the quality of service (QoS) of computer systems and networks is... more Performance Modelling to predict the quality of service (QoS) of computer systems and networks is widely applied. Similarly there is an abundant literature on the analysis and optimisation of supply chains and inventory control. This paper addresses some problems at the interface between these two areas, since computer systems and networks are widely used to control supply chains and E-Commerce systems. We develop an analytical modelling approach using queueing theory to study the impact of cyber-attacks on the economic performance of warehouse that sells perishable goods such as food that has a limited shelf life. We derive expressions to predict the loss of income that results from cyber-attacks on a web site used to order goods, and illustrate them with numerical examples.

NOLTA, IEICE An Energy Packet Network model for mobile networks with energy harvesting

Mobile communications are a powerful contributor to social and economic development worldwide, in... more Mobile communications are a powerful contributor to social and economic development worldwide, including in less developed or remote parts of the world. However they are large users of electricity through their base stations, backhaul networks and Cloud servers, so that they have a large environmental impact when they use the electric grid. On the other hand, they could operate with renewable energy sources and thus reduce their CO2 impact and be accessible even in areas where the electric grid is unavailable or unreliable. The counterpart is that intermittent sources of energy, such as photovoltaic and wind, can affect the quality of service (QoS) that is experienced by mobile users. Thus in this paper we model the performance of mobile telecommunications that use intermittent and renewable energy sources. In such cases to analyse the performance of such systems, both the energy supply and the network traffic, can be modeled as random processes, and we develop mathematical models using the Energy Packet Network paradigm, where both data and energy flows are discretised. QoS metrics for the users are computed based on the traffic intensity and the availability of energy.

Deep Learning with Dense Random Neural Networks for Detecting Attacks against IoT-connected Home Environments

by Erol Gelenbe and Dr Yasin Murat Kadioglu

In this paper, we analyze the network attacks that can be launched against IoT gateways, identify... more In this paper, we analyze the network attacks that can be launched against IoT gateways, identify the relevant metrics to detect them, and explain how they can be computed from packet captures. We also present the principles and design of a deep learning-based approach using dense random neural networks (RNN) for the online detection of network attacks. Empirical validation results on packet captures in which attacks were inserted show that the Dense RNN correctly detects attacks.

Current Research on Cybersecurity in Europe

These Proceedings [15] start with an overview [2] of the contents of this volume, providing insig... more These Proceedings [15] start with an overview [2] of the contents of this volume, providing insight into how some of these contributions are interconnected and linking them to prior ideas and work. It then follows with a series of research papers on Cyber- security research in Europe that covers five projects funded by the European Commis- sion:
– KONFIDO on the security of communications and data transfers for interconnected European national or regional health services,
– GHOST regarding the security of IoT systems for the home, and the design of secure IoT home gateways,
– SerIoT on the Cybersecurity of IoT systems in general with a range of applications in supply chains, smart cities, and other areas,
– NEMESYS, a now completed research project on the security of mobile networks, and SDK4ED, a new project that addresses security only incidentally but that focuses on broader issues of computation time, energy consumption and reliability of software.

GHOST -Safe-Guarding Home IoT Environments with Personalised Real-time Risk Control

We present the European research project GHOST, (Safeguarding home IoT environments with personal... more We present the European research project GHOST, (Safeguarding home IoT environments with personalised real-time risk control), which challenges the traditional cyber security solutions for the IoT by proposing a novel reference architecture that is embedded in an adequately adapted smart home network gateway, and designed to be vendor-independent. GHOST proposes to lead a paradigm shift in consumer cyber security by coupling usable security with transparency and behavioural engineering.

KONFIDO: An OpenNCP-based Secure eHealth Data Exchange System

by Erol Gelenbe and Evangelos Grivas

The European Commission is very focused on the development of solutions to allow effective cross-... more The European Commission is very focused on the development of solutions to allow effective cross-border health-care data interchange with the aim of guaranteeing a uniform QoS level of health-care systems across Europe. A first effort in this direction was made by the epSOS project with the OpenNCP platform which overcomes interoperability issues in patients health information exchange among European healthcare systems. However, some security issues are only partially solved, leading to the KONFIDO project which will address them by extending OpenNCP with a sound holistic approach to security at a sys-temic level. This paper describes the KONFIDO project's approach, discusses its design and its representation as a system of interacting agents. We also discuss how it is being deployed by combining complementary security enhancing technologies with the ultimate goal of increasing trust and security in data exchange systems for eHealth.

Chapter Title Deep Learning with Random Neural Networks Deep Learning with Random Neural Networks

by Erol Gelenbe and Yonghua Yin

This paper develops multi-layer classifiers and auto-encoders based on the Random Neural Network.... more This paper develops multi-layer classifiers and auto-encoders based on the Random Neural Network. Our motivation is to build robust classifiers that can be used in systems applications such as Cloud management for the accurate detection of states that can lead to failures. Using an idea concerning some to soma interactions between natural neuronal cells, we discuss a basic building block constructed of clusters of densely packet cells whose mathematical properties are based on G-Networks and the Random Neural Network. These mathematical properties lead to a transfer function that can be exploited for large arrays of cells. Based on this mathematical structure we build multi-layer networks. In order to evaluate the level of classification accuracy that can be achieved, we test these auto-encoders and classifiers on a widely used standard database of handwritten characters. Abstract. This paper develops multi-layer classifiers and auto-encoders based on the Random Neural Network. Our motivation is to build robust classifiers that can be used in systems applications such as Cloud management for the accurate detection of states that can lead to failures. Using an idea concerning some to soma interactions between natural neuronal cells, we discuss a basic building block constructed of clusters of densely packet cells whose mathematical properties are based on G-Networks and the Random Neural Network. These mathematical properties lead to a transfer function that can be exploited for large arrays of cells. Based on this mathematical structure we build multi-layer networks. In order to evaluate the level of classification accuracy that can be achieved, we test these auto-encoders and classifiers on a widely used standard database of handwritten characters. AQ1

Mobile Communications with Intermittent and Renewable Energy

—Mobile communications are a powerful contributor to social and economic development worldwide, a... more —Mobile communications are a powerful contributor to social and economic development worldwide, and in particular in less developed parts of the world. However the extension and penetration of mobile communications are often hampered by the state of the electrical grid, which may not cover all areas and which may also be intermittent and unreliable. Thus we review some of the economic and technological challenges for mobile telecommunications that integrate and exploit potentially plentiful renewable energy sources, such as photovoltaic, in developing areas of the world. Such sources of energy for communication networks are also useful to mitigate the environmental impact of energy consumption of Information and Communication Technologies (ICT) worldwide. However renewable energy sources are themselves intermittent, and this raises new challenges concerning network design. Hence this paper also develops an analytical approach that can be used to evaluate the quality of service of networks that operate with intermittent sources of energy.

Self-Aware Self-aware Computing Systems: Open Challenges and Future Research Directions

by Erol Gelenbe and Xiaoyun Zhu

In this chapter, we discuss the open challenges in building self-aware computing systems that are... more In this chapter, we discuss the open challenges in building self-aware computing systems that are still being faced by the research and development community. The challenges can be theoretical, technical, computational, or even sociological. First, we highlight the challenges associated with each of the earlier parts of the book and summarize on respective future research directions. We then offer concluding remarks and an outlook into the future in the last section.

Self Aware Networks

This article is a summary description of the cognitive packet network (CPN) which is an early exa... more This article is a summary description of the cognitive packet network (CPN) which is an early example of a completely software-defined network (SDN) and of a fully implemented self-aware computer network (SAN). CPN has been completely implemented and is used in numerous experiments. CPN is able to observe its own internal performance as well as the interfaces of the external systems that it interacts with, in order to modify its behaviour so as to adaptively achieve objectives, such as discovering services for its users, improving their quality of service (QoS), reducing its own energy consumption, compensating for components that fail or malfunction, detecting and reacting to intrusions, and defending itself against attacks. Abstract This article is a summary description of the cognitive packet network

Self-aware Networks: The Cognitive Packet Network and Its Performance

This article is a summary description of the cognitive packet network (CPN) which is an early exa... more This article is a summary description of the cognitive packet network (CPN) which is an early example of a completely software-defined network (SDN) and of a fully implemented self-aware computer network (SAN). CPN has been completely implemented and is used in numerous experiments. CPN is able to observe its own internal performance as well as the interfaces of the external systems that it interacts with, in order to modify its behaviour so as to adaptively achieve objectives, such as discovering services for its users, improving their quality of service (QoS), reducing its own energy consumption, compensating for components that fail or malfunction, detecting and reacting to intrusions, and defending itself against attacks. Abstract This article is a summary description of the cognitive packet network

Data Transaction Modeling in Mobile Networks: Contract Mechanism and Performance Analaysis

We consider auction mechanism design and performance analysis for data transactions in mobile soc... more We consider auction mechanism design and performance analysis for data transactions in mobile social networks. Existing mobile network plans can result in some users ending a monthly plan with excess data, while others may have to pay a costly fee to buy more data. Thus we suggest data auctions with a single seller, or a multiple-seller networked data auction, that operate in mobile social networks, to deal with the asymmetry between extra unused data resources and urgent data demands. Based on earlier work on the analysis of auctions, we design the data transaction mechanism, and summarise the analysis on state transmission, stationary probabilities of the system, and the expected income for data sellers. To improve the efficiency and performance of the system, a socially-aware mobility model is also proposed. The proposed data auction mechanisms and friendship-based mobility model are then simulated as operating on Flickr, a real-world online social network database. Results show that the number of data bidders in different auctions can be balanced through the proposed mobility model, and also increase the income per unit time of sellers in the networked data auction.

Deep Learning in Multi-Layer Architectures of Dense Nuclei

by Erol Gelenbe and Yonghua Yin

In dense clusters of neurons in nuclei, cells may interconnect via soma-to-soma interactions, in ... more In dense clusters of neurons in nuclei, cells may interconnect via soma-to-soma interactions, in addition to conventional synaptic connections. We illustrate this idea with a multi-layer architecture (MLA) composed of multiple clusters of recurrent sub-networks of spiking Random Neural Networks (RNN) with dense soma-to-soma interactions. We use this RNN-MLA architecture for deep learning. The inputs to the clusters are normalised by adjusting the external arrival rates of spikes to each cluster, and then apply this architectures to learning from multi-channel datasets. We present numerical results based on both images and sensor based data that show the value of this RNN-MLA for deep learning.