The pervasive game environments have activated explosive growth of the Internet over recent decad... more The pervasive game environments have activated explosive growth of the Internet over recent decades. Thus, understanding Internet traffic characteristics and precise classification have become important issues in network management, resource provisioning, and game application development. Naturally, much attention has been given to analyzing and modeling game traffic. Little research, however, has been undertaken on the classification of game traffic. In this paper, we perform an interpretive traffic analysis of popular game applications at the transport layer and propose a new classification method based on a simple decision tree, called an alternative decision tree (ADT), which utilizes the statistical traffic characteristics of game applications. Experimental results show that ADT precisely classifies game traffic from other application traffic types with limited traffic features and a small number of packets, while maintaining low complexity by utilizing a simple decision tree.
Time-division-multiplexed passive optical networks (TDM-PONs) have been successful solutions for ... more Time-division-multiplexed passive optical networks (TDM-PONs) have been successful solutions for providing the huge bandwidth needed for quadruple-play services in last-mile networks. In this paper, the effects of excess bandwidth provisioning applied to a reservation-based TDM are analyzed in detail, and noble adaptive excess timeslot provisioning (ETP) mechanisms are proposed in the interoptical network unit (ONU) and intra-ONU domain in the TDM-PONs. Using traffic load distribution information observed in an optical line termination (OLT), a proposed deficit reservation-based dynamic bandwidth allocation, called dr-DBA, controls the amounts of excess timeslot bandwidths in order to efficiently serve the burst traffic of heavy-loaded ONUs while guaranteeing strict fair link-sharing. The authors also introduce a service work-based ETP bandwidth reservation as well as an optimal reservation class selection mechanism to support prioritized service in an intra-ONU so that serious node light-load penalty and bandwidth overbooking can be minimized. The performance results achieved by numerical analysis and simulations prove that proposed ETP mechanisms provide fair and optimal excess bandwidths to the ONUs according to the changes in traffic load distribution, while compensating the systematic weakness of each mechanism in both ONU domains.
International Journal of Future Computer and Communication, 2013
Electronics and Telecommunication Research Institute (ETRI) and worked on the development of the ... more Electronics and Telecommunication Research Institute (ETRI) and worked on the development of the TDX digital switching system family, including TDX-1, TDX-1A, TDX-1B, TDX-10, and ATM switching systems. In 1998, he moved to Information and Communications University, Daejeon, South Korea as a member of faculty. Currently he is a professor of the Department of Electrical and Electronics Engineering, KAIST, Daejeon, South Korea. His research interests are network architecture, network protocols, and performance analysis of telecommunication systems. He is a member of the IEEE, IEEK, and KICS of South Korea.
2013 Fifth International Conference on Ubiquitous and Future Networks (ICUFN), 2013
Recently, the Peer-to-Peer (P2P) technology has been emerging as a promising technology for QoS i... more Recently, the Peer-to-Peer (P2P) technology has been emerging as a promising technology for QoS issued services especially in delivering video contents (e.g., video on demand (VoD), live streaming services, etc). Although traffic of P2P streaming services has been much increased, little information on their traffic behaviors or workload is available on research community. In this paper, for the best of our knowledge, we first present traffic characteristics of one of popular live personal broadcasting services based upon P2P, namely Afreeca, in Korea, which supports various types of video qualities. Based upon the deep packet inspection and comprehensive analysis at packet and flow levels, we illustrate traffic characteristics in regard to streaming rates, distributions pertaining to packet sizes and inter-arrival times, session duration, and popularity of the broadcasting jockeys (BJs), etc with several numerical models exploiting goodness of fitting test.
Adaptive CAC algorithms using measurement and estimation
Proceedings IEEE International Conference on Communications ICC '95, 1995
To prevent network congestion due to unpredictable statistical fluctuation of broadband traffic p... more To prevent network congestion due to unpredictable statistical fluctuation of broadband traffic pattern, a robust and reliable traffic control mechanism is required in an asynchronous transfer mode (ATM) network. We propose new adaptive connection admission control (CAC) algorithms which calculate the cell loss performance of related function blocks in a switch node using only a declared peak cell rate and an estimated mean cell rate to reduce the inaccuracy of the declared source parameters. We measure only the mean cell rate of the aggregate cell stream in a link to guarantee measurement accuracy. The proposed CAC algorithms can allow more relaxed source parameter declarations and thus guarantee efficient bandwidth utilization
International Conference on Networking, International Conference on Systems and International Conference on Mobile Communications and Learning Technologies (ICNICONSMCL'06), 2006
Each network has its own distinct quality of service (QoS) class. To support end-to-end QoS QoS c... more Each network has its own distinct quality of service (QoS) class. To support end-to-end QoS QoS class interwoking between different network technologies is very important function especially in Next Generation Networks (NGN). QoS mapping between different networks technologies are broadly categorized as mapping between parameters and mapping between classes in each different networks. However, these methods have some problems that the the former is too complicated to implement or the latter may execute uncetrain mapping between networks because of granularity QoS class. So in this paper we present two QoS class mapping methodologies over heterogenous networks to provide end-to-end QoS support for application services. And we propose a QoS class mapping method in different transport techonologies mixed circumstance using Application Service Map (ASM) which classifies application services based on performance requirements. Ths mapping method is based on location information of ASM to determine QoS class according to underlying techonology. The proposed mapping method can easy to accept new coming application services without alternation or modification and it can provide users with the more refined QoS.
16th International Conference on Advanced Communication Technology, 2014
The problem of minimizing the number of coding nodes is caused by network coding overhead and is ... more The problem of minimizing the number of coding nodes is caused by network coding overhead and is proved to be NP-hard. To resolve this issue, this paper proposes Centralitybased Network Coding Node Selection (CNCNS) that is the heuristic and distributed mechanism to minimize the number of network coding (NC) nodes without compromising the achievable network throughput. CNCNS iteratively analyses the node centrality and selects NC node in the specific area. Since CNCNS operates with distributed manner, it can dynamically adapt the network status with approximately minimizing network coding nodes. Especially, CNCNS adjusts the network performance of network throughput and reliability using control indicator. Simulation results show that the well selected network coding nodes can improve the network throughput and almost close to throughput performance of a system where all network nodes operate network coding.
16th International Conference on Advanced Communication Technology, 2014
This paper explores the trade-off relationship between a spatial diversity gain and a control ove... more This paper explores the trade-off relationship between a spatial diversity gain and a control overhead with regard to the number of hop count of the routing tree in WiMAX mesh networks. The effective gain or overhead with the number of hop count is influenced by the density of SSs and traffic load. So this paper proposes the algorithm adjusting the depth of routing tree according to the time varying number of SSs and traffic load for achieving optimal throughput performance. The simulation results show the proposed algorithm can have throughput performance enhancement in the time varying traffic load environment.
Network coding, whose intermediate nodes decode and encode packets to forward incoming packets, i... more Network coding, whose intermediate nodes decode and encode packets to forward incoming packets, increases the capacity the multicast network. However there are security issues of confidentiality and integrity which degrade and threat network coding. In this paper, we propose Secure Key Checksum (SKC), a secure mechanism using an encryption key, to provide both confidentiality and integrity. To provide the confidentiality, SKC utilizes a block encryption. For the integrity check, SKC uses a checksum based scheme verifying encoded blocks on-the-fly with low computational cost. Also, SKC uses a combined security key for encryption and checksum.
A QoS-Based Adaptive Resource Sharing Protection for Optical Burst Switching Networks
Information Networking. Advances in Data Communications and Wireless Networks, 2006
Abstract. Protection and restoration are essential mechanisms for guaranteeing more reliable traf... more Abstract. Protection and restoration are essential mechanisms for guaranteeing more reliable traffic delivery services. But it is not easy to apply existing mechanisms to optical burst switching (OBS) networks due to its one-way reservation signaling and the statistical burst multi- ...
A New Size-Based Burst Assembly Scheme for OBS Switches
Lecture Notes in Computer Science, 2006
We propose a new size-based burst classification algorithm which assembles arriving packets with ... more We propose a new size-based burst classification algorithm which assembles arriving packets with different burst size according to classes in OBS ingress switches. In JET-based OBS scheme, because the high priority class bursts are likely to occupy channels in advance, the low priority class bursts should find void intervals between high-class bursts to avoid burst blocking. The research results show that there was close relation between blocking rate of low-class bursts and size of its bursts. By limiting the maximum length of low-class bursts, loss, throughput and assembly delay performances are improved when proposed scheme is compared to conventional scheme.
Test and performance comparison of end-to-end available bandwidth measurement tools
The bandwidth measurement is essential for resource management, provisioning of network. There ar... more The bandwidth measurement is essential for resource management, provisioning of network. There are many measurements techniques and tools, but the accuracy and intrusiveness of these tools are not much shown concurrently in a paper. Thus, we survey bandwidth measurement tools and evaluate performance of some well-known tools. In this paper, we present measurement results in accuracy and intrusiveness perspectives with IGI-PTR, pathChirp. Also, we present traffic volume of probing packets and measurement time using probing packet inspection with tcpdump. The test results give us information about what we should consider when we measure bandwidth.
In this paper, we propose flow level service classification according to application's deter... more In this paper, we propose flow level service classification according to application's deterministic characteristics in packet size and flow length, and design service differentiation scheme in best-effort networks with packet size and flow length based random early detection (PF-RED). In our results, PF-RED can provide service differentiation without sacrificing overall performance.
This paper describes restoration techniques for OBS networks. First, we introduce the design issu... more This paper describes restoration techniques for OBS networks. First, we introduce the design issues to be essentially considered, when restoration techniques will be built into OBS networks. Second, considering the design issues, we propose a novel restoration model named "Recover-Then-Reserve," which can enhance the restoration performance. Third, we develop a simulator using the NS-2 platform to verify the performance of the restoration techniques (path, link, and sub-path) designed by each restoration model (GMPLS-based OBS restoration and our proposal). Performance is evaluated on the following metrics: restoration success rate, restoration time, connection availability, and burst loss probability. The effects of the failure frequency and the number of the restoration attempts on the restoration performance are also investigated. Our experiments show that the proposal can reduce the restoration time by about 25 ms compared to the GMPLS-based OBS restoration model. Also, path restoration technique has better performance than link and sub-path restoration techniques because it has the highest restoration success rate, allowing similar restoration time. Our experiments also show that the more the restoration attempts made and the less the failure frequency becomes, the better the restoration performance obtained.
Enhanced Class-of-Service Oriented Packet Scheduling Scheme for EPON Access Networks
IEICE Transactions on Communications, 2008
A fast class-of-service oriented packet scheduling (FCOPS) has a service fairness problem since a... more A fast class-of-service oriented packet scheduling (FCOPS) has a service fairness problem since a credit pool for a service class is initialized at the beginning of a transmission cycle whose starting moment is fixed at a specific ONU. To remedy the service unfairness of FCOPS, we suggest an enhanced class-of-service oriented packet scheduling (ECOPS) that uses a new initialization cycle
This paper presents a novel collision-less optical burst transmission scheme, namely, look-ahead ... more This paper presents a novel collision-less optical burst transmission scheme, namely, look-ahead optical burst transmission (LAOBT) for WDM ring networks. In the LAOBT, bursts are generated and inserted into burst streams by utilizing the void intervals of incoming bursts (IBs) in advance. The void size is estimated by splitting optical power into two paths, delaying one of the signals at the fiber delay line (FDL) and calculating IBs' residual time from the splitter to the optical cross-connect (OXC). The performance evaluation results show that our proposed scheme increases link utilization without significant increment of end-to-end delay and signaling overhead.
The pervasive game environments have activated explosive growth of the Internet over recent decad... more The pervasive game environments have activated explosive growth of the Internet over recent decades. Thus, understanding Internet traffic characteristics and precise classification have become important issues in network management, resource provisioning, and game application development. Naturally, much attention has been given to analyzing and modeling game traffic. Little research, however, has been undertaken on the classification of game traffic. In this paper, we perform an interpretive traffic analysis of popular game applications at the transport layer and propose a new classification method based on a simple decision tree, called an alternative decision tree (ADT), which utilizes the statistical traffic characteristics of game applications. Experimental results show that ADT precisely classifies game traffic from other application traffic types with limited traffic features and a small number of packets, while maintaining low complexity by utilizing a simple decision tree.
Time-division-multiplexed passive optical networks (TDM-PONs) have been successful solutions for ... more Time-division-multiplexed passive optical networks (TDM-PONs) have been successful solutions for providing the huge bandwidth needed for quadruple-play services in last-mile networks. In this paper, the effects of excess bandwidth provisioning applied to a reservation-based TDM are analyzed in detail, and noble adaptive excess timeslot provisioning (ETP) mechanisms are proposed in the interoptical network unit (ONU) and intra-ONU domain in the TDM-PONs. Using traffic load distribution information observed in an optical line termination (OLT), a proposed deficit reservation-based dynamic bandwidth allocation, called dr-DBA, controls the amounts of excess timeslot bandwidths in order to efficiently serve the burst traffic of heavy-loaded ONUs while guaranteeing strict fair link-sharing. The authors also introduce a service work-based ETP bandwidth reservation as well as an optimal reservation class selection mechanism to support prioritized service in an intra-ONU so that serious node light-load penalty and bandwidth overbooking can be minimized. The performance results achieved by numerical analysis and simulations prove that proposed ETP mechanisms provide fair and optimal excess bandwidths to the ONUs according to the changes in traffic load distribution, while compensating the systematic weakness of each mechanism in both ONU domains.
International Journal of Future Computer and Communication, 2013
Electronics and Telecommunication Research Institute (ETRI) and worked on the development of the ... more Electronics and Telecommunication Research Institute (ETRI) and worked on the development of the TDX digital switching system family, including TDX-1, TDX-1A, TDX-1B, TDX-10, and ATM switching systems. In 1998, he moved to Information and Communications University, Daejeon, South Korea as a member of faculty. Currently he is a professor of the Department of Electrical and Electronics Engineering, KAIST, Daejeon, South Korea. His research interests are network architecture, network protocols, and performance analysis of telecommunication systems. He is a member of the IEEE, IEEK, and KICS of South Korea.
2013 Fifth International Conference on Ubiquitous and Future Networks (ICUFN), 2013
Recently, the Peer-to-Peer (P2P) technology has been emerging as a promising technology for QoS i... more Recently, the Peer-to-Peer (P2P) technology has been emerging as a promising technology for QoS issued services especially in delivering video contents (e.g., video on demand (VoD), live streaming services, etc). Although traffic of P2P streaming services has been much increased, little information on their traffic behaviors or workload is available on research community. In this paper, for the best of our knowledge, we first present traffic characteristics of one of popular live personal broadcasting services based upon P2P, namely Afreeca, in Korea, which supports various types of video qualities. Based upon the deep packet inspection and comprehensive analysis at packet and flow levels, we illustrate traffic characteristics in regard to streaming rates, distributions pertaining to packet sizes and inter-arrival times, session duration, and popularity of the broadcasting jockeys (BJs), etc with several numerical models exploiting goodness of fitting test.
Adaptive CAC algorithms using measurement and estimation
Proceedings IEEE International Conference on Communications ICC '95, 1995
To prevent network congestion due to unpredictable statistical fluctuation of broadband traffic p... more To prevent network congestion due to unpredictable statistical fluctuation of broadband traffic pattern, a robust and reliable traffic control mechanism is required in an asynchronous transfer mode (ATM) network. We propose new adaptive connection admission control (CAC) algorithms which calculate the cell loss performance of related function blocks in a switch node using only a declared peak cell rate and an estimated mean cell rate to reduce the inaccuracy of the declared source parameters. We measure only the mean cell rate of the aggregate cell stream in a link to guarantee measurement accuracy. The proposed CAC algorithms can allow more relaxed source parameter declarations and thus guarantee efficient bandwidth utilization
International Conference on Networking, International Conference on Systems and International Conference on Mobile Communications and Learning Technologies (ICNICONSMCL'06), 2006
Each network has its own distinct quality of service (QoS) class. To support end-to-end QoS QoS c... more Each network has its own distinct quality of service (QoS) class. To support end-to-end QoS QoS class interwoking between different network technologies is very important function especially in Next Generation Networks (NGN). QoS mapping between different networks technologies are broadly categorized as mapping between parameters and mapping between classes in each different networks. However, these methods have some problems that the the former is too complicated to implement or the latter may execute uncetrain mapping between networks because of granularity QoS class. So in this paper we present two QoS class mapping methodologies over heterogenous networks to provide end-to-end QoS support for application services. And we propose a QoS class mapping method in different transport techonologies mixed circumstance using Application Service Map (ASM) which classifies application services based on performance requirements. Ths mapping method is based on location information of ASM to determine QoS class according to underlying techonology. The proposed mapping method can easy to accept new coming application services without alternation or modification and it can provide users with the more refined QoS.
16th International Conference on Advanced Communication Technology, 2014
The problem of minimizing the number of coding nodes is caused by network coding overhead and is ... more The problem of minimizing the number of coding nodes is caused by network coding overhead and is proved to be NP-hard. To resolve this issue, this paper proposes Centralitybased Network Coding Node Selection (CNCNS) that is the heuristic and distributed mechanism to minimize the number of network coding (NC) nodes without compromising the achievable network throughput. CNCNS iteratively analyses the node centrality and selects NC node in the specific area. Since CNCNS operates with distributed manner, it can dynamically adapt the network status with approximately minimizing network coding nodes. Especially, CNCNS adjusts the network performance of network throughput and reliability using control indicator. Simulation results show that the well selected network coding nodes can improve the network throughput and almost close to throughput performance of a system where all network nodes operate network coding.
16th International Conference on Advanced Communication Technology, 2014
This paper explores the trade-off relationship between a spatial diversity gain and a control ove... more This paper explores the trade-off relationship between a spatial diversity gain and a control overhead with regard to the number of hop count of the routing tree in WiMAX mesh networks. The effective gain or overhead with the number of hop count is influenced by the density of SSs and traffic load. So this paper proposes the algorithm adjusting the depth of routing tree according to the time varying number of SSs and traffic load for achieving optimal throughput performance. The simulation results show the proposed algorithm can have throughput performance enhancement in the time varying traffic load environment.
Network coding, whose intermediate nodes decode and encode packets to forward incoming packets, i... more Network coding, whose intermediate nodes decode and encode packets to forward incoming packets, increases the capacity the multicast network. However there are security issues of confidentiality and integrity which degrade and threat network coding. In this paper, we propose Secure Key Checksum (SKC), a secure mechanism using an encryption key, to provide both confidentiality and integrity. To provide the confidentiality, SKC utilizes a block encryption. For the integrity check, SKC uses a checksum based scheme verifying encoded blocks on-the-fly with low computational cost. Also, SKC uses a combined security key for encryption and checksum.
A QoS-Based Adaptive Resource Sharing Protection for Optical Burst Switching Networks
Information Networking. Advances in Data Communications and Wireless Networks, 2006
Abstract. Protection and restoration are essential mechanisms for guaranteeing more reliable traf... more Abstract. Protection and restoration are essential mechanisms for guaranteeing more reliable traffic delivery services. But it is not easy to apply existing mechanisms to optical burst switching (OBS) networks due to its one-way reservation signaling and the statistical burst multi- ...
A New Size-Based Burst Assembly Scheme for OBS Switches
Lecture Notes in Computer Science, 2006
We propose a new size-based burst classification algorithm which assembles arriving packets with ... more We propose a new size-based burst classification algorithm which assembles arriving packets with different burst size according to classes in OBS ingress switches. In JET-based OBS scheme, because the high priority class bursts are likely to occupy channels in advance, the low priority class bursts should find void intervals between high-class bursts to avoid burst blocking. The research results show that there was close relation between blocking rate of low-class bursts and size of its bursts. By limiting the maximum length of low-class bursts, loss, throughput and assembly delay performances are improved when proposed scheme is compared to conventional scheme.
Test and performance comparison of end-to-end available bandwidth measurement tools
The bandwidth measurement is essential for resource management, provisioning of network. There ar... more The bandwidth measurement is essential for resource management, provisioning of network. There are many measurements techniques and tools, but the accuracy and intrusiveness of these tools are not much shown concurrently in a paper. Thus, we survey bandwidth measurement tools and evaluate performance of some well-known tools. In this paper, we present measurement results in accuracy and intrusiveness perspectives with IGI-PTR, pathChirp. Also, we present traffic volume of probing packets and measurement time using probing packet inspection with tcpdump. The test results give us information about what we should consider when we measure bandwidth.
In this paper, we propose flow level service classification according to application's deter... more In this paper, we propose flow level service classification according to application's deterministic characteristics in packet size and flow length, and design service differentiation scheme in best-effort networks with packet size and flow length based random early detection (PF-RED). In our results, PF-RED can provide service differentiation without sacrificing overall performance.
This paper describes restoration techniques for OBS networks. First, we introduce the design issu... more This paper describes restoration techniques for OBS networks. First, we introduce the design issues to be essentially considered, when restoration techniques will be built into OBS networks. Second, considering the design issues, we propose a novel restoration model named "Recover-Then-Reserve," which can enhance the restoration performance. Third, we develop a simulator using the NS-2 platform to verify the performance of the restoration techniques (path, link, and sub-path) designed by each restoration model (GMPLS-based OBS restoration and our proposal). Performance is evaluated on the following metrics: restoration success rate, restoration time, connection availability, and burst loss probability. The effects of the failure frequency and the number of the restoration attempts on the restoration performance are also investigated. Our experiments show that the proposal can reduce the restoration time by about 25 ms compared to the GMPLS-based OBS restoration model. Also, path restoration technique has better performance than link and sub-path restoration techniques because it has the highest restoration success rate, allowing similar restoration time. Our experiments also show that the more the restoration attempts made and the less the failure frequency becomes, the better the restoration performance obtained.
Enhanced Class-of-Service Oriented Packet Scheduling Scheme for EPON Access Networks
IEICE Transactions on Communications, 2008
A fast class-of-service oriented packet scheduling (FCOPS) has a service fairness problem since a... more A fast class-of-service oriented packet scheduling (FCOPS) has a service fairness problem since a credit pool for a service class is initialized at the beginning of a transmission cycle whose starting moment is fixed at a specific ONU. To remedy the service unfairness of FCOPS, we suggest an enhanced class-of-service oriented packet scheduling (ECOPS) that uses a new initialization cycle
This paper presents a novel collision-less optical burst transmission scheme, namely, look-ahead ... more This paper presents a novel collision-less optical burst transmission scheme, namely, look-ahead optical burst transmission (LAOBT) for WDM ring networks. In the LAOBT, bursts are generated and inserted into burst streams by utilizing the void intervals of incoming bursts (IBs) in advance. The void size is estimated by splitting optical power into two paths, delaying one of the signals at the fiber delay line (FDL) and calculating IBs' residual time from the splitter to the optical cross-connect (OXC). The performance evaluation results show that our proposed scheme increases link utilization without significant increment of end-to-end delay and signaling overhead.
Uploads
Papers by Hong Shik Park