With recent improvements in wireless communication technology, the proportion of high-loss and high-latency wireless links such as satellite communication links has increased. Conventional TCP communication based on the end-to-end principle suffers from significant throughput degradation owing to the inefficiency of retransmission control when relatively high loss rates and high-latency wireless links are included. To solve this problem, we previously proposed a dynamic splitting TCP method called TCP-DOPPEL+ (TCP-Dynamically Opt for Proxy to achieve Performance Enhancement based on Link state PLUS), which is based on the product of the packet loss rate and latency. However, TCP-DOPPEL+ uses the product of the packet loss rate and latency as the splitting threshold. If one of the measured values deviates from the actual situation, it becomes difficult to place the splitting devices appropriately. Therefore, we propose a dynamic splitting TCP method, TCP-HORUS (TCP-proxy Hack and Optimize Retransmission by Utilizing link State), based on the packet loss rate and latency with parameters used independently. To evaluate the characteristics of TCP-HORUS, we conducted experiments to evaluate the throughput environments where links with different packet loss rates and latencies between links are mixed on the route, and their locations are varied. The results show that TCP-HORUS improves throughput by at most 20.6% compared to TCP-DOPPEL+.

As the use of information technology (IT) in the business operations of organizations such as companies is progressing, and the online services of the organizations become increasingly prevalent in employees of companies, etc., threats such as abuse of online services by third parties and attacks on the network and servers have emerged. To reduce the risk of such threats, it is necessary to estimate the degree of the trustworthiness of users who have accessed online service accounts and to understand what threats exist for each individual user’s access. Furthermore, to grasp the overall picture of threats to the network and online services, it would be useful and effective to estimate the percentage of users who have accessed online service accounts in a certain period with the intent to attack. This paper proposes and evaluates a method to estimate the trustworthiness of a group of users who have accessed online service accounts in a certain period as the percentage of accesses from the users associated with their accounts among all accesses that include those with the intent to attack. The method also uses only information on each user’s behavior observable while the network is operational, apart from the authentication results. The results show that the proposed method can estimate the percentage of accesses from the users associated with their accounts with an error of a few percent and be effective at controlling the strength of security measures, including the strength of authentication, in the operation of the network and online services, depending on the increase or decrease in the percentage of accesses from the users associated with their accounts.

In smart grids, interdependency between a communication network and a power network causes a risk that a physical failure within the power network invokes cascading failures spreading over a wide range of the smart grid. Although backup power units prepared for communication nodes can prevent expansion of the cascading failures, their capacity may be insufficient for their operation during entire recovery period. Hence, timing of initiating the backup power units is crucial to minimize a negative impact of the cascading failures over the entire recovery period. This paper formulates a backup power scheduling problem for optimizing the timing to put each backup power unit into operation when damaged power nodes are recovered progressively through multiple recovery-subperiods. Hybrid greedy method based on heuristics related to utilization efficiency of the backup power units is proposed to derive a suboptimal solution from restricted information on the progress recovery. A lightweight method is also proposed to derive a backup power schedule for each recovery-subperiod individually. Greedy methods incorporating the proposed lightweight method can derive approximated solutions of the backup power scheduling problem in polynomial time. The effectiveness of the proposed methods is verified through simulation experiments.

Ku-band BS broadcasting signal (11.84 GHz, circular polarization, elevation angle 41.3°) and Ka-band CS beacon signal (19.45 GHz, circular polarization, elevation angle 49.5°) have been measured at Osaka Electro-Communication University (Neyagawa, Osaka) from 1988 to 2006. A frequency scaling method from Ku-band to Ka-band rain attenuation is investigated, taking into account the effects of raindrop size distribution (DSD). Actual measurements of the kind of DSD by an optical sensor show good agreement with the theoretical value of rain attenuation ratio of CS to BS radio waves. Based on this method, the rain attenuation that occurs in higher frequency bands such as 30 GHz and 40 GHz is estimated. The rain attenuation estimates for 30 GHz and 40 GHz show similar results between the frequency scaling methods using both Ku-band BS and Ka-band CS attenuation. Long-term statistics for the 19 years from 1988 to 2006, however, do not show so significant differences, whether the DSD is selected for each rainfall event or designated uniformly. After 2007, the rain attenuation statistics of 30 and 40 GHz are successfully estimated by the same frequency scaling method from only Ku-band BS attenuation. According to the increase of BS rain attenuation statistics, those of 30 GHz and 40 GHz are increased about 1.4 times for the time percentages of 0.01-0.2% after 2007.

This paper proposes low complexity resource allocation based on a linear search for frequency domain non-orthogonal multiple access based on the low-density signature (LDS). Two algorithms are proposed for the linear search. The application of the linear search makes the proposed resource allocation implemented with lower and fixed computational complexity. The performance of the non-orthogonal access based on the linear search is evaluated by computer simulation. One of the proposed linear search algorithm makes the non-orthogonal multiple access achieve 5 dB better BER performance than the other when the overloading ratio is set to 2. The complexity of the non-orthogonal access based on the proposed linear search algorithm is approximately half as much as that of the conventional low complexity resource allocation when the overloading ratio is 2, if the complexity is evaluated in terms of the number of additions.