Papers by Nattapong Kitsuwan
IEICE Technical Report; IEICE Tech. Rep., Oct 19, 2017
The Journal of Institute of Electronics, Information and Communication Engineers, Sep 1, 2019
This paper proposes a switch migration scheme to balance the load of all controllers in multiple ... more This paper proposes a switch migration scheme to balance the load of all controllers in multiple controllers based software-defined network. Several works define the number of requested packets as a controller load. The controller needs to hold the current processing until receiving a response message from the switch before processing a new request. For this reason, the number of requested packets incorporates with the path cost is defined as a controller load. In a multiple controllers approach, a load imbalanced state is recognized if a load of one controller is biased against a load of the other controllers. In the imbalanced state, connections of some switches that belong to a controller with high load are disconnected, and the switches are reconnected to the other controllers with the lower load. This process is called a migration process. The probability of migration and the efficiency of migration are used as a decision to find a switch to be migrated and the other controller to be reconnected. The load imbalanced state is resolved but the aggregate load of all controllers may be increased. In the proposed scheme, the aggregate load of all controllers after the migration is used as an indicator to select the migrated switch and a new controller. A pair of migrated switch and controller with the minimum aggregate load is selected so that the aggregate load is reduced compared to the conventional scheme. Computer simulation result shows that the proposed scheme reduces the aggregate load, compared to the traditional scheme.
IEICE Technical Report; IEICE Tech. Rep., Oct 30, 2012
The Journal of Engineering, 2009
An implementation technique for traffic splitting using a meter table of a standard OpenFlow in a... more An implementation technique for traffic splitting using a meter table of a standard OpenFlow in a software-defined network is proposed. The proposed technique uses different differentiated services code point (DSCP) numbers, which are generally used to classify network traffic for quality of service (QoS) levels, in a packet header to decide an output port for a corresponding route. The meter table is generally designed for a QoS purpose. The OpenFlow standard provides two types of meter band, 'drop' and 'dscp_remark'. The proposed technique adopts the 'dscp_remark' type for a splitting traffic purpose, instead of the QoS purpose. The DSCP number of the packet is modified when the packet rate exceeds a pre-defined traffic rate. Packets with different DSCP numbers are sent out to the neighbour switches from their corresponding output ports. The original DSCP number is returned to the packets when they are sending out to the neighbour switches, so that the QoS is not affected.
2023 International Conference on Information Networking (ICOIN)
2016 IEEE Conference on Standards for Communications and Networking (CSCN)
A switch in a software-defined network (SDN) uses a flow table for packet forwarding. The flow ta... more A switch in a software-defined network (SDN) uses a flow table for packet forwarding. The flow table consists of flow entries decided by a controller. As user requests with a variety of data types have been dramatically increasing, a controller's CPU is overloaded to process those requests. The users may not obtain real-time data. Furthermore, the control plane might be overflowed by OpenFlow messages that correspond to the requests. The number of requests can be reduced by a scheme to keep permanent flow entries in the switch. A huge number of permanent flow entries to support the forwarding function degrade other functions. This paper proposes a scheme to reduce the number of permanent flow entries and the number of OpenFlow messages generated by the controller. The proposed scheme divides switches in the network into several regional groups. It re-designs a flow table that is incorporated with a concept of two multiprotocol label switching tags to guide packets from sources to destinations. One tag guides the packets from a source switch to an edge switch in the destination group. The other tag guides the packets from the edge switch to another switch in the local group. Performance evaluation shows 74% reductions of the number of permanent flow entries, compared to a conventional scheme that keeps permanent flow entries without inquiry to the controller. Moreover, 81% of the number of OpenFlow messages is reduced, compared to another scheme that inquires the controller every time once a request arrives at the switch.
The Journal of Institute of Electronics, Information and Communication Engineers, Sep 1, 2019
2019 19th International Symposium on Communications and Information Technologies (ISCIT), 2019
This paper proposes an algorithm to resolve the imbalance state in a software-defined network in ... more This paper proposes an algorithm to resolve the imbalance state in a software-defined network in which orchestrated multiple-controller architecture is applied. In the orchestrated multiple-controller architecture, a super-controller recognizes load of all controllers in the network and run a load balancing process. Controllers process flow requests from switches based on the decision of the super-controller. The load of a controller in this paper includes a propagation delay and the number of packet-in messages, since the controller needs to wait for a response from the instruction requested switch for a flow installation. The load balancing is a technique to efficiently use the resources, to increase throughput, to reduce the response time for a request, and to avoid the imbalance state of all controllers. When the imbalance state is detected by the super-controller, requests of a switch in a controller with the maximum load are transferred to process at another available controller. Load of the maximum controller is reduced. A load balancing algorithm to pair switch and controller is introduced to minimize the maximum load of all controllers. Computer simulation result shows that the proposed algorithm reduces up to 79.11% of the average standard deviation, which indicates the imbalance state, of all controllers, compared to a conventional scheme.
The Journal of Institute of Electronics, Information and Communication Engineers, Sep 1, 2019
2019 International Conference on Computing, Networking and Communications (ICNC), 2019
This paper proposes a spectrum allocation scheme to reduce the number of blocked requests for an ... more This paper proposes a spectrum allocation scheme to reduce the number of blocked requests for an elastic optical network (EON) in which an original spectrum band is allowed be sliced. In a general EON, a spectrum allocation must not conflict two constraints. First, the allocated spectrum band must be continuous from a source to a destination if no spectrum converter is applied. Second, spectrum slots for the allocated spectrum band must be consecutive in a link. Recently, a technology to slice and to recover the spectrum band has been invented. The second constraint of the general EON can be broken. The proposed scheme applies a slicing technology in the EON. An algorithm to allocate the spectrum is introduced. Results show that the request blocking of the proposed scheme reduces up to 82% and 13% in a low traffic and high traffic, respectively, compared to the general EON.
2017 19th International Conference on Transparent Optical Networks (ICTON), 2017
This paper evaluates the performance of the first-last-exact fit spectrum allocation policy for e... more This paper evaluates the performance of the first-last-exact fit spectrum allocation policy for elastic optical networks. The first-last-exact fit spectrum allocation policy is intended to increase the number of aligned available slots and avoid small contiguous available slots, and hence suppresses the blocking probability. This policy separates the allocation of disjoint and non-disjoint connections. Connections with disjoint paths are allocated using the first-exact fit allocation policy, whereas connections with non-disjoint paths are allocated using the last-exact fit allocation policy. The separation of the disjoint and non-disjoint connections provides a higher number of aligned available slots. Simulation results show that the first-last-exact fit spectrum allocation policy improves the traffic admissibility compared to the conventional first fit and first-last fit policies.
2019 19th International Symposium on Communications and Information Technologies (ISCIT), 2019
Denial of service (DoS) is a process of injecting malicious packets into the network. Intrusion d... more Denial of service (DoS) is a process of injecting malicious packets into the network. Intrusion detection system (IDS) is a system used to investigate malicious packets in the network. Software-defined network (SDN) physically separates control plane and data plane. The control plane is moved to a centralized controller, and it makes a decision in the network from a global view. The combination between IDS and SDN allows the prevention of malicious packets to be more efficient due to the advantage of the global view in SDN. IDS needs to communicate with switches to have an access to all end-to-end traffic in the network. The high traffic in the link between switches and IDS results in congestion. The congestion between switches and IDS delays the detection and prevention of malicious traffic. To address this problem, we propose a historical database (Hdb), a scheme to reduce the traffic between switches and IDS, based on the historical information of a sender. The simulation shows that in the average, 54.1% of traffic mirrored to IDS is reduced compared to the conventional schemes.
2021 International Conference on Information Networking (ICOIN), 2021
A slicing and stitching technology has been invented to relax the consecutive constraint of spect... more A slicing and stitching technology has been invented to relax the consecutive constraint of spectrum slot allocation in elastic optical network (EON). This technology splits a spectrum band into several signal bands, called optical components, by making a copy of the original spectrum band and filtering out an unwanted signal on each spectrum band. The remaining optical components are injected into a transmission channel. At the destination, the optical components are recovered by using phase preserving wavelength conversion. Therefore, a request is able to allocate to dis-consecutive groups of slots. A conventional spectrum allocation scheme with this technology adopts slicing devices, called spectrum slicers, at only a source node. There is a problem of allocation patterns due to lack of flexibility since slicing at intermediate nodes is not considered. In this paper, we propose a spectrum allocation scheme considering slicers at both source node and intermediate nodes. Performance of the proposed scheme is evaluated by a computer simulation. The results show that the proposed scheme with 20 slicers reduces 45% of request blocking probability compared to the conventional scheme in COST239 topology when the traffic is 300 Erlang.
IEICE technical report. Speech, 2009
Hadoop has swiftly been accepted in both industry and academia in order to offer enormous storage... more Hadoop has swiftly been accepted in both industry and academia in order to offer enormous storage for any kind of data, massive processing power and the capability to manage virtually jobs. This paper designs a Hadoop system, which considers both slave server’s processing capacity and network delay for wide area networks in order to reduce the job processing time. The demonstrated testbed results designate that the task allocation scheme in our designed Hadoop system is effective.
Uploads
Papers by Nattapong Kitsuwan