論文

DoS 攻撃や DDoS 攻撃に対し，その発信元を特定するIP トレースバックは重要な技術である．これまで，マーキング方式やロギング方式が個別に提案されてきたが，マーキング方式ではトレースバックのために大量の攻撃パケットの回収が必要であるという課題があり，またロギング方式では全ての通過パケットのハッシュ値を保持するためルータへの負荷が増大するという課題がある．近年，２つの方式の課題を補う HIT(Hybrid IP Traceback) 方式が提案された．HIT方式ではマーキングとロギングを交互に行うことでロギング回数を従来と比較し半分に低減しているが，全ての通過パケットに対して処理を行わなければならない課題や，DDoS攻撃において被害ホスト近傍のルータへ大きな負荷がかかるといった課題がある．そこで本論文では， DDoS 攻撃に対するトレースバックにおいて， 95% 以上の高いトレースバック成功率を達成しつつルータへの負荷を低減・分散するために，排他的論理和と確率的パケットマーキングを用いたトレースバック方式を提案する．排他的論理和を用いることで 2 つのルータの ID を 1 つにまとめることができ，ロギングの回数を低減することが可能となる．また， DDoS 攻撃において攻撃パケットは大量に送信されてくるという特徴と被害ホスト近傍に攻撃パケットが集中するということを考慮して，全ての通過パケットではなく確率的に選択したパケットに対して 1 度のみ処理を行う．これにより，ルータへの負荷を低減することが可能となる．計算機シミュレーションにより各方式におけるトレースバック成功率とパケット処理率，またパケット処理回数の評価を行い，高いトレースバック成功率を達成しつつルータへの負荷を低減・分散可能な提案方式の有効性を示す．

We propose a primary traffic based multihop routing with the cooperative transmission(PTBMR-CT) algorithm that enlarges the hop transmission distances to reduce the number of hops of the route from the cognitive source (CS) to the cognitive destination (CD). In each hop, from the cognitive nodes in a specified area depending on the status of the primary traffic, the cognitive node that is farthest away from the cognitive sender is selected as the cognitive receiver. However, when the primary source (PS) is transmitting the data to the primary destination (PD), another cognitive node is also selected to be the cognitive receiver for the cooperative transmission from the cognitive nodes that are in a specified area. The cooperative transmission is performed if the PS still transmits the data to the PD when the cognitive receiver of the next hop is being searched. The simulation results show that the average number of hops is reduced by the PTBMR-CT compared to the conventional primary traffic based farthest neighbor routing (PTBFNR), and the conventional PTBFNR is outperformed by the PTBMR-CT in terms of the average end-to-end reliability, the average end-to-end throughput, and the average required transmission power of transmitting the data from the CS to the CD.

We propose a grid-based routing protocol which divides the network area for large side length of square cells by using cell rotation to reduce the number of relay nodes between the source node and the mobile sink. The proposed scheme divides each cell into multiple sub-cells, and assumes one or two sub-cells to be active-cells. Then, the proposed scheme confines the existing area of active nodes to each active-cell. Because the maximum transmission range is a fixed value, the side length of square cells can be enlarged by confining the area where the active node exist in each cell. Therefore, the proposed scheme decreases the number of relay nodes between the source node and the mobile sink due to the large cells divided, and reduces the data packets delivery latency and the energy consumption. The proposed scheme introduces three types of dividing method, which assume 2-, 4-, and 6-division method. The high numbers of division method improves each characteristic values due to expanding the side length of square cells under the condition of high density of sensor nodes. The simulation results show that the proposed scheme improves the packets latency and the energy consumption efficiency.

In Mobile IPv6(MIPv6), the Mobile Node(MN) communicating with the Correspondent Node(CN) cannot prove ownership of the claimed IP address. If a malicious node impersonates victim's IP address, it could hijack the session or forward packets to non-existing destination or other nodes. Currently, in order for the MN to prove ownership of its own IP address, it is considered using Feige-Fiat-Shamir (FFS) Identification Scheme cite{FFS}. However, there is one serious problem. In FFS Identification Scheme, the CN has to verify all of the Binding Update requests, and this leads to DoS(Denial of Service) attack. This paper shows a method which mitigates the effect of DoS attack by making Challenge twice in one transaction. We make the first Challenge easy to verify it to exclude the malicious nodes and the second one much more difficult than the first one to avoid impersonation. This method can efficiently exclude malicious nodes which do not have proper IP addresses by verifying the first Challenge. Furthermore, by making the Challenge twice, our scheme can decrease the probability of impersonation over the conventional scheme for almost the same amount of calculation. By computer simulations, we show that the proposed scheme is efficient to decrease the effect of the DoS attack and the probability of impersonation compared to the conventional scheme.

We propose the asynchronous receiver-initiated MAC protocol with the stair-like sleep that each node reduces its own sleep time by the sleep-change-rate depending on the number of hops from the source to the sink in wireless sensor networks (WSNs). Using the stair-like sleep, our protocol achieves the high delivery ratio, the low packet delay, and the high energy efficiency due to the reduction of the idle listening time. Our protocol can formulate the upper bound of the idle listening time because of the feature that the sleep time decreases in the geometric progression, and the reduction of the idle listening time is obtained by using the stair-like sleep. In our proposed scheme, the sink calculates the sleep change rate based on the number of hops from the source to the sink. By using the control packets which have the role of ACK, our proposed protocol can achieve the stair-like sleep by no additional control packets. In addition, even in the network condition that multi-targets are detected, and hops to the sink are changed frequently, our proposed protocol can change the sleep change rate adaptively because the sink can always obtain the number of hops from the source to the sink. Simulation results show that the proposed protocol can improve the performances in the packet delivery ratio, the packet delay, and the energy efficiency compared to the conventional receiver-initiated MAC (RI-MAC) protocol.

Light-Weight energy efficiency Media Access Control (LWMAC) is an asynchronous duty cycle MAC protocol to shorten the preamble length by selecting the node which is changed to be in active mode first during the preamble as a next hop node without selecting it in advance, and then reduces energy consumption in wireless sensor networks. However, LWMAC may increase the number of hops because it doesn't consider the distance from a source node to a next hop node. Therefore, when a source node selects a node which is far to a sink node as a next hop node, energy consumption cannot be reduced. In this paper, we propose a duty cycle MAC protocol that decreases the number of hops and reduces energy consumption by selecting a node that is closer to a sink node as a next hop node when more than two candidate nodes change to be in active mode until the end of preamble whose length is same in LWMAC. By the simulation, we evaluate the average number of hops of LWMAC and the proposed scheme. The simulation results show that the proposed scheme reduces the energy consumption by decreasing the average number of hops.

Tracing IP packets back to their origins is an important scheme to defend against Denial Of Service(DoS) and distributed DoS (DDoS) attacks. Mainly, two kinds of IP Traceback schemes have been proposed. One is a packet marking scheme that each router marks their ID into packets. The other is a packet logging scheme that each router logs the digest of the forwarded packets. Recently, hybrid IP Traceback scheme, which makes up for shortcomings of above two schemes, has been proposed. In this hybrid scheme, the overhead on routers can be reduced compared to the conventional scheme. However, due to marking and logging for DDoS attacks, the hybrid scheme still causes much overhead, and routers near victim server have lots of burden. In this paper, we propose an IP Traceback scheme which uses exclusive-or and probabilistic packet marking to reduce burden on routers while achieving high traceback success rate compared to the conventional schemes. The proposed scheme uses not only packet marking and packet logging but also exclusive-or. Because two routers ID can be combined by using exclusive-or, the number of routers which log packets decreases. Furthermore, considering that huge number of packets gather around routers near a victim server for DDoS attacks, each router makes its job(marking, logging, exclusive-or) probabilistically and only once. By taking this scheme, the closer routers are to a victim server, the smaller probability that routers make its job is, and it can reduce burden on each router. By computer simulation, we evaluate a traceback success rate, a probability that each router make its job to forwarding packets and packet numbers that each router makes its job. We show that our proposed scheme can achieve high traceback success rate while reducing huge burden on routers compared to conventional schemes.

The performance of multi-user MIMO-OFDM Multiple Access Channel (MIMO-OFDM-MAC) Uplink systems can be significantly increased by using adaptive transmission and efficient resource-allocation scheme in the system. In the MIMO-OFDM-MAC scenario, the base station decides centrally on the optimal bit loading and subcarrier allocation as well as transmit antenna selection from multiple users based on channel state information (CSI) in the system. There have a lot of papers which have done on the resource allocation in MIMO-OFDM system. But most of them consider in MIMO-OFDM downlink and very few papers consider to give the fairness among users. Moreover, there has no paper which considers to give the proportional data rate fairness among user in the uplink MIMO-OFDM-MAC system. In this paper, resource allocation for spatial and spectral dimension and adaptive bit loading scheme is proposed to improve the total system capacity under the constraint of total transmit power and target bit error rate (BER) for each user and proportional data rate fairness among users in the uplink MIMO-OFDM-MAC system.

We propose a novel multihop cooperative routing (MCR) algorithm that extends the lengths of the x axis projections of the hop distances and finds the multihop route that minimizes the number of hops in spectrum sharing networks. At each hop except the last one, firstly, the cognitive relay that is nearest from the cognitive sender is selected. If the participation of the selected cognitive relay can not satisfy the QoS requirement of the cognitive transmission, among the cognitive receiver candidates that satisfy the QoS requirement of the cognitive transmission, the one whose x coordinate has the largest difference from that of the cognitive relay is selected as the cognitive receiver. At the last hop, if the cognitive relay is available, the cooperative transmission is performed. Otherwise, the direct transmission is performed. Simulation results show that the MCR reduces the average number of hops compared to the FNR and outperfoms the FNR in terms of the average end-to-end reliability, the average end-toend throughput, and the average required transmission power.