Search results for: vehicular ad hoc network

Authors: Amarpreet Singh, Kimi Manchanda

Abstract:

Intended for providing the security in the VANETS (Vehicular Ad hoc Network) scenario, the covert storage channel is implemented through data transmitted between the sender and the receiver. Covert channels are the logical links which are used for the communication purpose and hiding the secure data from the intruders. This paper refers to the Establishment of bit selective mode covert storage channels in VANETS. In this scenario, the data is being transmitted with two modes i.e. the normal mode and the covert mode. During the communication between vehicles in this scenario, the controlling of bits is possible through the optional bits of IPV6 Header Format. This implementation is fulfilled with the help of Network simulator.

Keywords: covert mode, normal mode, VANET, OBU, on-board unit

Procedia PDF Downloads 338

Authors: Ibrahim Obeidat

Abstract:

Reaching a fully-connected network of mobile nodes in rural areas got a great attention between network researchers. This attention rose due to the complexity and high costs while setting up the needed infrastructures for these networks, in addition to the low transmission range these nodes has. Terranet technology, as an example, employs ad-hoc mesh network where each node has a transmission range not exceed one kilometer, this means that every two nodes are able to communicate with each other if they are just one kilometer far from each other, otherwise a third-party will play the role of the “relay”. In Terranet, and as an idea to reduce network setup cost, every node in the network will be considered as a router that is responsible of forwarding data between other nodes which result in a decentralized collaborative environment. Most researches on Terranet presents the idea of how to encourage mobile nodes to become more cooperative by letting their devices in “ON” state as long as possible while accepting to play the role of relay (router). This research presents the issue of finding the percentage of nodes in ad-hoc mesh network within rural areas that should play the role of relay at every time slot, relating to what is the actual area coverage of nodes in order to have the network reach the fully-connectivity. Far from our knowledge, till now there is no current researches discussed this issue. The research is done by making an implementation that depends on building adjacency matrix as an indicator to the connectivity between network members. This matrix is continually updated until each value in it refers to the number of hubs that should be followed to reach from one node to another. After repeating the algorithm on different area sizes, different coverage percentages for each size, and different relay percentages for several times, results extracted shows that for area coverage less than 5% we need to have 40% of the nodes to be relays, where 10% percentage is enough for areas with node coverage greater than 5%.

Keywords: ad-hoc mesh networks, network connectivity, mobile ad-hoc networks, Terranet, adjacency matrix, simulator, wireless sensor networks, peer to peer networks, vehicular Ad hoc networks, relay

Procedia PDF Downloads 245

Authors: C. Benjbara, A. Habbani

Abstract:

Wireless networks are getting more and more used in every new technology or feature, especially those without infrastructure (Ad hoc mode) which provide a low cost alternative to the infrastructure mode wireless networks and a great flexibility for application domains such as environmental monitoring, smart cities, precision agriculture, and so on. These application domains present a common characteristic which is the need of coexistence and intercommunication between modules belonging to different types of ad hoc networks like wireless sensor networks, mesh networks, mobile ad hoc networks, vehicular ad hoc networks, etc. This vision to bring to life such heterogeneous networks will make humanity duties easier but its development path is full of challenges. One of these challenges is the communication complexity between its components due to the lack of common or compatible protocols standard. This article proposes a new patented routing protocol based on the OLSR standard in order to resolve the heterogeneous ad hoc networks communication issue. This new protocol is applied on a specific network architecture composed of MANET, VANET, and FANET.

Keywords: Ad hoc, heterogeneous, ID-Node, OLSR

Procedia PDF Downloads 178

Authors: Rashi Jain

Abstract:

The Wireless Body Area Networks (WBANs) is an emerging interdisciplinary area where small sensors are placed on/within the human body. These sensors monitor the physiological activities and vital statistics of the body. The data from these sensors is aggregated and communicated to a remote doctor for immediate attention or to a database for records. On 6 Feb 2012, the IEEE 802.15.6 task group approved the standard for Body Area Network (BAN) technologies. The standard proposes the physical and MAC layer for the WBANs. The work provides an introduction to WBANs and overview of the physical and MAC layers of the standard. The physical layer specifications have been covered. A comparison of different protocols used at MAC layer is drawn. An introduction to the network layer and security aspects of the WBANs is made. The WBANs suffer certain limitations such as regulation of frequency bands, minimizing the effect of transmission and reception of electromagnetic signals on the human body, maintaining the energy efficiency among others. This has slowed down their implementation.

Keywords: vehicular networks, sensors, MicroController 8085, LTE

Procedia PDF Downloads 233

Authors: Subhadip Biswas, Shivendra Maurya, Satish Chandra, Indrajit Ghosh

Abstract:

Use of speed prediction models sometimes appears as a feasible alternative to laborious field measurement particularly, in case when field data cannot fulfill designer’s requirements. However, developing speed models is a challenging task specifically in the context of developing countries like India where vehicles with diverse static and dynamic characteristics use the same right of way without any segregation. Here the traffic composition plays a significant role in determining the vehicular speed. The present research was carried out to examine the effects of traffic volume and its composition on vehicular speed under mixed traffic conditions. Classified traffic volume and speed data were collected from different geometrically identical six lane divided arterials in New Delhi. Based on these field data, speed prediction models were developed for individual vehicle category adopting Kriging approximation technique, an alternative for commonly used regression. These models are validated with the data set kept aside earlier for validation purpose. The predicted speeds showed a great deal of agreement with the observed values and also the model outperforms all other existing speed models. Finally, the proposed models were utilized to evaluate the effect of traffic volume and its composition on speed.

Keywords: speed, Kriging, arterial, traffic volume

Procedia PDF Downloads 330

Authors: Farshad Zeinali, Sajedeh Norouzi, Nader Mokari, Eduard Jorswieck

Abstract:

The capacity of fifth-generation (5G) vehicle-to-everything (V2X) networks poses significant challenges. To ad- dress this challenge, this paper utilizes New Radio (NR) and New Radio Unlicensed (NR-U) networks to develop a heterogeneous vehicular network (HetNet). We propose a new framework, named joint BS assignment and resource allocation (JBSRA) for mobile V2X users and also consider coexistence schemes based on flexible duty cycle (DC) mechanism for unlicensed bands. Our objective is to maximize the average throughput of vehicles while guaranteeing the WiFi users' throughput. In simulations based on deep reinforcement learning (DRL) algorithms such as deep deterministic policy gradient (DDPG) and deep Q network (DQN), our proposed framework outperforms existing solutions that rely on fixed DC or schemes without consideration of unlicensed bands.

Keywords: vehicle-to-everything (V2X), resource allocation, BS assignment, new radio (NR), new radio unlicensed (NR-U), coexistence NR-U and WiFi, deep deterministic policy gradient (DDPG), deep Q-network (DQN), joint BS assignment and resource allocation (JBSRA), duty cycle mechanism

Procedia PDF Downloads 68

Authors: Auxeeliya Jesudoss, Ashraph Sulaiman, Ratnakar Kotnana

Abstract:

V2V communication brings up several questions of scalability issues although message sharing in vehicular ad-hoc networks comprises of both Vehicle-to-Vehicle communications (V2V) and Vehicle to Infrastructure communication (V2I). It is not an easy task for a vehicle to verify all signatures of the messages sent by its neighboring vehicles in a timely manner, without resulting in message loss. Moreover, the communication overhead of a vehicle to authenticate another vehicle would increase together with the security of the system. Another issue to be addressed is the continuous mobility of vehicles which requires at least some information on the node’s own position to be revealed to the neighboring vehicles. This may facilitate the attacker to congregate information on a node’s position or its mobility patterns. In order to tackle these issues, this paper introduces a RSU aggregated message deliverance scheme called RAMeD. With RAMeD, roadside units (RSUs) are responsible for verifying the identity of the vehicles entering in its range, collect messages from genuine vehicles and to aggregate similar messages into groups before sending them to all the vehicles in its communication range. This aggregation will tremendously improve the rate of message delivery and reduce the message lose ratio by avoiding similar messages being sent to the vehicles redundantly. The proposed protocol is analyzed extensively to evaluate its merits and efficiency for vehicular communication.

Keywords: vehicular ad-hoc networks, V2V, V2I, VANET communication, scalability, message aggregation

Procedia PDF Downloads 375

Authors: Markus Mazzola, Gunther Schaaf

Abstract:

A vehicle driving with an Adaptive Cruise Control System (ACC) is usually controlled decentrally, based on the information of radar systems and in some publications based on C2X-Communication (CACC) to guarantee stable platoons. In this paper, we present a Model Predictive Control (MPC) design of a centralized, server-based ACC-System, whereby the vehicular platoon is modeled and controlled as a whole. It is then proven that the proposed MPC design guarantees asymptotic stability and hence string stability of the platoon. The Networked MPC design is chosen to be able to integrate system constraints optimally as well as to reduce the effects of communication delay and packet loss. The performance of the proposed controller is then simulated and analyzed in an LTE communication scenario using the LTE/EPC Network Simulator LENA, which is based on the ns-3 network simulator.

Keywords: adaptive cruise control, centralized server, networked model predictive control, string stability

Procedia PDF Downloads 484

Authors: Ranbir Singh, Deepinder Kaur

Abstract:

Network security is an important aspect in every field like government offices, Educational Institute and any business organization. Network security consists of the policies adopted to prevent and monitor forbidden access, misuse, modification, or denial of a computer network. Network security is very complicated subject and deal by only well trained and experienced people. However, as more and more people become wired, an increasing number of people need to understand the basics of security in a networked world. The history of the network security included an introduction to the TCP/IP and interworking. Network security starts with authenticating, commonly with a username and a password. In this paper, we study about various types of attacks on network security and how to handle or prevent this attack.

Keywords: network security, attacks, denial, authenticating

Procedia PDF Downloads 370

Authors: Salman Naseer

Abstract:

One of the main challenges of operating a smart city (SC) is collecting the massive data generated from multiple data sources (DS) and to transmit them to the control units (CU) for further data processing and analysis. These ever-increasing data demands require not only more and more capacity of the transmission channels but also results in resource over-provision to meet the resilience requirements, thus the unavoidable waste because of the data fluctuations throughout the day. In addition, the high energy consumption (EC) and carbon discharges from these data transmissions posing serious issues to the environment we live in. Therefore, to overcome the issues of intensive EC and carbon emissions (CE) of massive data dissemination in Smart Cities, we propose an energy efficient and carbon reduction approach by utilizing the daily mobility of the existing vehicles as an alternative communications channel to accommodate the data dissemination in smart cities. To illustrate the effectiveness and efficiency of our approach, we take the Auckland City in New Zealand as an example, assuming massive data generated by various sources geographically scattered throughout the Auckland region to the control centres located in city centre. The numerical results show that our proposed approach can provide up to 5 times lower delay as transferring the large volume of data by utilizing the existing daily vehicles’ mobility than the conventional transmission network. Moreover, our proposed approach offers about 30% less EC and CE than that of conventional network transmission approach.

Keywords: smart city, delay tolerant network, infrastructure offloading, opportunistic network, vehicular mobility, energy consumption, carbon emission

Procedia PDF Downloads 114

Authors: Thomas Adi Purnomo Sidhi

Abstract:

Nowadays, network is an essential need in almost every part of human daily activities. People now can seamlessly connect to others through the Internet. With advanced technology, our personal data now can be more easily accessed. One of many components we are concerned for delivering the best network is a security issue. This paper is proposing a method that provides more options for security. This research aims to improve network security by focusing on the physical layer which is the first layer of the OSI model. The layer consists of the basic networking hardware transmission technologies of a network. With the use of observation method, the research produces a schematic design for enhancing the network security through the gray code converter.

Keywords: network, network security, grey code, physical layer

Procedia PDF Downloads 470

Authors: M. Magrini, D. Moroni, G. Palazzese, G. Pieri, D. Azzarelli, A. Spada, L. Fanucci, O. Salvetti

Abstract:

Railway crossings are complex entities whose optimal management cannot be addressed unless with the help of an intelligent transportation system integrating information both on train and vehicular flows. In this paper, we propose an integrated system named SIMPLE (Railway Safety and Infrastructure for Mobility applied at level crossings) that, while providing unparalleled safety in railway level crossings, collects data on rail and road traffic and provides value-added services to citizens and commuters. Such services include for example alerts, via variable message signs to drivers and suggestions for alternative routes, towards a more sustainable, eco-friendly and efficient urban mobility. To achieve these goals, SIMPLE is organized as a System of Systems (SoS), with a modular architecture whose components range from specially-designed radar sensors for obstacle detection to smart ETSI M2M-compliant camera networks for urban traffic monitoring. Computational unit for performing forecast according to adaptive models of train and vehicular traffic are also included. The proposed system has been tested and validated during an extensive trial held in the mid-sized Italian town of Montecatini, a paradigmatic case where the rail network is inextricably linked with the fabric of the city. Results of the tests are reported and discussed.

Keywords: Intelligent Transportation Systems (ITS), railway, railroad crossing, smart camera networks, radar obstacle detection, real-time traffic optimization, IoT, ETSI M2M, transport safety

Procedia PDF Downloads 475

Authors: Kenichiro Hida, Shin-Ichi Kuribayashi

Abstract:

NFV-based network implements a variety of network functions with software on general-purpose servers, and this allows the network operator to select any capabilities and locations of network functions without any physical constraints. In this paper, we evaluate the influence of the maximum tolerable network delay on the virtual routing function deployment guidelines which the authors proposed previously. Our evaluation results have revealed the following: (1) the more the maximum tolerable network delay condition becomes severe, the more the number of areas where the route selection function is installed increases and the total network cost increases, (2) the higher the routing function cost relative to the circuit bandwidth cost, the increase ratio of total network cost becomes larger according to the maximum tolerable network delay condition.

Keywords: NFV (Network Functions Virtualization), resource allocation, virtual routing function, minimum total network cost

Procedia PDF Downloads 215

Authors: Amel Ltifi, Ahmed Zouinkhi, Mohamed Salim Bouhlel

Abstract:

Security and trust management in Vehicular Ad-hoc NETworks (VANET) is a crucial research domain which is the scope of many researches and domains. Although, the majority of the proposed trust management systems for VANET are based on specific road infrastructure, which may not be present in all the roads. Therefore, road security should be managed by vehicles themselves. In this paper, we propose a new Self Organized Trust Management system (SOTM). This system has the responsibility to cut with the spread of false warnings in the network through four principal components: cooperation, trust management, communication and security.

Keywords: ative vehicle, cooperation, trust management, VANET

Procedia PDF Downloads 397

Authors: Nihed Bahria El Asghar, Imen Jouili, Mounir Frikha

Abstract:

Inter-datacenter transport network is very bandwidth and delay demanding. The data transferred over such a network is also highly QoS-exigent mostly because a huge volume of data should be transported transparently with regard to the application user. To avoid the data transfer failure, a backup path should be reserved. No re-routing delay should be observed. A dedicated 1+1 protection is however not applicable in inter-datacenter transport network because of the huge spare capacity. In this context, we propose a survivable virtual network with minimal backup based on network coding (1 ⊕ 1) and solve it using a modified Dijkstra-based heuristic.

Keywords: network coding, dedicated protection, spare capacity, inter-datacenters transport network

Procedia PDF Downloads 419

Authors: Takumi Shindo, Koji Okamura

Abstract:

The first generation of the wide area network was circuit centric network. How the optimal circuit can be signed was the most important issue to get the best performance. This architecture had succeeded for line based telephone system. The second generation was host centric network and Internet based on this architecture has very succeeded world widely. And Internet became as new social infrastructure. Currently the architecture of the network is based on the location of the information. This future network is called Information centric network (ICN). The information-centric network (ICN) has being researched by many projects and different architectures for implementation of ICN have been proposed. The goal of this study is to compare performances of those ICN architectures. In this paper, the authors propose general ICN model which can represent two typical ICN architectures and compare communication performances using request routing. Finally, simulation results are shown. Also, we assume that this network architecture should be adapt to energy on-demand routing.

Keywords: ICN, information centric network, CCN, energy

Procedia PDF Downloads 304

Authors: David Tena-Gago, Jose M. Alcaraz Calero, Qi Wang

Abstract:

The evolution and integration of automated vehicles have become more and more tangible in recent years. State-of-the-art technological advances in the field of camera-based Artificial Intelligence (AI) and computer vision greatly favor the performance and reliability of the Advanced Driver Assistance System (ADAS), leading to a greater knowledge of vehicular operation and resembling human behavior. However, the exclusive use of this technology still seems insufficient to control vehicular operation at 100%. To reveal the degree of accuracy of the current camera-based automated driving AI modules, this paper studies the structure and behavior of one of the main solutions in a controlled testing environment. The results obtained clearly outline the lack of reliability when using exclusively the AI model in the perception stage, thereby entailing using additional complementary sensors to improve its safety and performance.

Keywords: accuracy assessment, AI-driven mobility, artificial intelligence, automated vehicles

Procedia PDF Downloads 81

Authors: Syed Umair Aziz, Muhammad Faheem, Sameer Hussain, Faraz Idris

Abstract:

Content centric network is the network based on the mechanism of sending and receiving the data based on the interest and data request to the specified node (which has cached data). In this network, the security is bind with the content not with the host hence making it host independent and secure. In this network security is applied by taking content’s MAC (message authentication code) and encrypting it with the public key of the receiver. On the receiver end, the message is first verified and after verification message is saved and decrypted using the receiver's private key.

Keywords: content centric network, client-server, host security threats, message authentication code, named data network, network caching, peer-to-peer

Procedia PDF Downloads 613

Authors: Kimaya Subhash Gaikwad, S. B. Waykar

Abstract:

The mobile ad-hoc network (MANET) is collection of various types of nodes. In MANET various protocols are used for communication. In OLSR protocol, a node is selected as multipoint relay (MPR) node which broadcast the messages. As the MANET is open kind of network any malicious node can easily enter into the network and affect the performance of the network. The performance of network mainly depends on the components which are taking part into the communication. If the proper nodes are not selected for the communication then the probability of network being attacked is more. Therefore, it is important to select the more reliable and secure components in the network. MANET does not have any filtering so that only selected nodes can be used for communication. The openness of the MANET makes it easier to attack the communication. The most of the attack are on the Quality of service (QoS) of the network. This paper gives the overview of the various attacks that are possible on OLSR protocol and some solutions. The papers focus mainly on the OLSR protocol.

Keywords: communication, MANET, OLSR, QoS

Procedia PDF Downloads 419

Authors: Maath M. Musleh

Abstract:

This research aims to study recent trends in the Palestinian feminist movement through the case study of Tal3at. The study uses social network analysis as its primary method to analyze Twitter data. It attempts to interpret results through the lens of network theories and Parson’s AGIL paradigm. The study reveals major structural weaknesses in the Tal3at network. Our findings suggest that the movement will decline soon as sentiments of alienation amongst Palestinian women increases. These findings were validated by a couple of central actors in the network. This study contributes an SNA approach to the understanding of the understudied Palestinian feminism.

Keywords: feminism, Palestine, social network analysis, Tal3at

Procedia PDF Downloads 220

Authors: Abdulqader Rasool Feqi Mohammed, Ergun Erçelebi̇

Abstract:

A large number of hospitals from developed countries are adopting hospital information system to bring efficiency in hospital information system. The purpose of this project is to research on new network security techniques in order to enhance the current network security structure of save a hospital information system (HIS). This is very important because, it will avoid the system from suffering any attack. Security architecture was optimized but there are need to keep researching on best means to protect the network from future attacks. In this final project research, security techniques were uncovered to produce best network security results when implemented in an integrated framework.

Keywords: hospital information system, HIS, network security techniques, internet protocol, IP, network

Procedia PDF Downloads 397

Authors: Ahmed Jedidi, Mesfer Mohammed Alshamrani, Alwi Mohammad A. Bamhdi

Abstract:

Optical performance monitoring and optical network management are essential in building a reliable, high-capacity, and service-differentiation enabled all-optical network. One of the serious problems in this network is the fact that optical crosstalk is additive, and thus the aggregate effect of crosstalk over a whole AON may be more nefarious than a single point of crosstalk. As results, we note a huge degradation of the Quality of Service (QoS) in our network. For that, it is necessary to identify and monitor the impairments in whole network. In this way, this paper presents new system to identify and monitor crosstalk in AONs in real-time fashion. particular, it proposes a new technique to manage intra-crosstalk in objective to relax QoS of the network.

Keywords: all-optical networks, optical crosstalk, optical cross-connect, crosstalk, monitoring crosstalk

Procedia PDF Downloads 426

Authors: Sidra Shabbir, Ayesha Manzoor, Mehreen Sirshar

Abstract:

The use of the network communication has imposed serious threats to the security of assets over the network. Network security is getting more prone to active and passive attacks which may result in serious consequences to data integrity, confidentiality and availability. Various cryptographic techniques have been proposed in the past few years to combat with the concerned problem by ensuring quality but in order to have a fully secured network; a framework of new cryptosystem was needed. This paper discusses certain cryptographic techniques which have shown far better improvement in the network security with enhanced quality assurance. The scope of this research paper is to cover the security pitfalls in the current systems and their possible solutions based on the new cryptosystems. The development of new cryptosystem framework has paved a new way to the widespread network communications with enhanced quality in network security.

Keywords: cryptography, network security, encryption, decryption, integrity, confidentiality, security algorithms, elliptic curve cryptography

Procedia PDF Downloads 702

Authors: Feng-jie Xie, Jian-hong Yan

Abstract:

Based on the complex network theory, we construct the air cargo network of the Belt and Road area, analyze its regional distribution and structural characteristics, measure the robustness of the network. The regional distribution results show that Southeast Asia and China have the most prominent development in the air cargo network of the Belt and Road area, Central Asia is the least developed. The structure characteristics found that the air cargo network has obvious small-world characteristics; the degree distribution has single-scale property; it shows a significant rich-club phenomenon simultaneously. The network robustness is measured by two attack strategies of degree and betweenness, but the betweenness of network nodes has a greater impact on network connectivity. And identified 24 key cities that have a large impact on the robustness of the network under the two attack strategies. Based on these results, recommendations are given to maintain the air cargo network connectivity in the Belt and Road area.

Keywords: air cargo, complex network, robustness, structure properties, The Belt and Road

Procedia PDF Downloads 165

Authors: Jin Xu

Abstract:

5G network developers need to satisfy the necessary requirements of additional capacity from massive users and spectrally efficient wireless technologies. Therefore, the significant amount of underutilized spectrum in network is motivating operators to combine long-term evolution (LTE) with intelligent spectrum management technology. This new LTE intelligent spectrum management in unlicensed band (LTE-U) has the physical layer topology to access spectrum, specifically the 5-GHz band. We proposed a new intelligent cloud RAN for 5G.

Keywords: cloud radio access network, wireless network, cloud computing, multi-agent

Procedia PDF Downloads 394

Authors: V. Andal Priyadharshini, Anumalasetty Yashwanth Nath

Abstract:

Network automation has evolved into a solution that ensures efficiency in all areas. The age-old technique to configure common software-defined networking protocols is inefficient as it requires a box-by-box approach that needs to be repeated often and is prone to manual errors. Network automation assists network administrators in automating and verifying the protocol configuration to ensure consistent configurations. This paper implemented network automation using Python and Ansible to configure different protocols and configurations in the container lab virtual environment. Ansible can help network administrators minimize human mistakes, reduce time consumption, and enable device visibility across the network environment.

Keywords: Python network automation, Ansible configuration, container lab deployment, software-defined networking, networking lab

Procedia PDF Downloads 133

Authors: Nashrawan Naser Taha, Andrew M. Cox

Abstract:

In a multi-cultural learning context, where ties are weak and dynamic, combining qualitative with quantitative research methods may be more effective. Such a combination may also allow us to answer different types of question, such as about people’s perception of the network. In this study the use of observation, interviews and photos were explored as ways of enhancing data from social network questionnaires. Integrating all of these methods was found to enhance the quality of data collected and its accuracy, also providing a richer story of the network dynamics and the factors that shaped these changes over time.

Keywords: mixed methods, social network analysis, multi-cultural learning, social network dynamics

Procedia PDF Downloads 487

Authors: Mohsen Dourandish

Abstract:

Iran has a huge pipeline network in every state of country which is the longest and vastest pipeline network after Russia and USA (360,000 Km high pressure pipelines and 250,000 Km distribution networks). Furthermore in recent years National Iranian Gas Company is planning to develop natural gas network to cover all cities and villages above 20 families, in a way that 97 percent of Iran population will be gas consumer by 2020. In this condition, network resiliency will be the first priority of NIGC and due to that several planning for increasing resiliency of gas network is under construction. The most important strategy of NIGC is converting tree form pattern network to loop gas networks and developing underground gas storage near main gas consuming centers. In this regard NIGC is planning for construction of over 3500 km high-pressure pipeline and also 10 TCM gas storage capacities in UGSs.

Keywords: Iranian gas network, peak shaving, resiliency, underground gas storage

Procedia PDF Downloads 296

Authors: Motonobu Hattori

Abstract:

In neural networks, when new patters are learned by a network, they radically interfere with previously stored patterns. This drawback is called catastrophic forgetting. We have already proposed a biologically inspired dual-network memory model which can much reduce this forgetting for static patterns. In this model, information is first stored in the hippocampal network, and thereafter, it is transferred to the neocortical network using pseudo patterns. Because, temporal sequence learning is more important than static pattern learning in the real world, in this study, we improve our conventional dual-network memory model so that it can deal with temporal sequences without catastrophic forgetting. The computer simulation results show the effectiveness of the proposed dual-network memory model.

Keywords: catastrophic forgetting, dual-network, temporal sequences, hippocampal

Procedia PDF Downloads 239

Authors: Min Jindong, Wang Mingxia

Abstract:

With the widespread use of artificial intelligence in life, computer vision, especially deep convolutional neural network models, has developed rapidly. With the increase of the complexity of the real visual target detection task and the improvement of the recognition accuracy, the target detection network model is also very large. The huge deep neural network model is not conducive to deployment on edge devices with limited resources, and the timeliness of network model inference is poor. In this paper, knowledge distillation is used to compress the huge and complex deep neural network model, and the knowledge contained in the complex network model is comprehensively transferred to another lightweight network model. Different from traditional knowledge distillation methods, we propose a novel knowledge distillation that incorporates multi-faceted features, called M-KD. In this paper, when training and optimizing the deep neural network model for target detection, the knowledge of the soft target output of the teacher network in knowledge distillation, the relationship between the layers of the teacher network and the feature attention map of the hidden layer of the teacher network are transferred to the student network as all knowledge. in the model. At the same time, we also introduce an intermediate transition layer, that is, an intermediate guidance layer, between the teacher network and the student network to make up for the huge difference between the teacher network and the student network. Finally, this paper adds an exploration module to the traditional knowledge distillation teacher-student network model. The student network model not only inherits the knowledge of the teacher network but also explores some new knowledge and characteristics. Comprehensive experiments in this paper using different distillation parameter configurations across multiple datasets and convolutional neural network models demonstrate that our proposed new network model achieves substantial improvements in speed and accuracy performance.

Keywords: object detection, knowledge distillation, convolutional network, model compression

Procedia PDF Downloads 248