In this article, we propose a new protection scheme for backbone networks to guarantee high service availability. The presented scheme does not require any reconfiguration immediately after the failure (i.e., it is proactive). At the same time, it does not require any reserved backup network resources either. To achieve these seemingly contradictory goals, we utilize the recent advancements in Machine Learning (ML) to implement a network intelligence that periodically re-allocates the unused capacity as protection bandwidth to meet the service availability requirements of each connection. Our goal is achieved by two components (1) predicting the traffic for the next period on each link, and (2) intelligently selecting the best fit dedicated protection scheme for the next period depending on the estimated unused (spare) bandwidth and the previous service availability violations. Note that re-allocating protection bandwidth affects neither the operational connections nor the current best practice of operators to over-provision network bandwidth to support elephant flows. Finally, we provide a case study on the real traffic from Energy Sciences Network (ESnet), a high-speed, international scientific backbone network. The key benefit of our framework is that adaptively utilizing the over-provisioned bandwidth for spare capacity is sufficient to improve the availability from three-nines to five-nines (in ESnet for the 30 examined connections). The drawback is negligible bandwidth limitations; the user perceives a minor and very temporal bandwidth limitation in less than 0.1% of the time.

The paper is concerned with a novel adaptive game server protocol optimization to combat network latencies in the case of heterogeneous network environment. In this way, game playing becomes feasible for clients accessing the game via different networks, which can pave the way to securing a higher income for the game industry and service providers. The sever based game protocol is viewed as a number of arrival processes from the clients (which are characterized by different delays) and periodical updates sent by the server to the clients. Game quality is quantified by two measures: (i) the tail probability of the maximal idle time; and (ii) the probability of missing an update period. Our objective is to chose server update time subject to minimizing the probability of the maximal idle time exceeding a certain threshold (which threshold is associated with the "psycho-physically approved quality of the game"). In order to calculate the optimal server update time and the underlying tail distribution, statistical tools from large deviation theory are used. Furthermore, an adaptive on-line algorithm has been developed which can adjust the server update time by estimating the corresponding delay probability density functions based on past observations. Due to the new method, game quality can be significantly improved despite the wide range of client latencies which typically characterize the heterogeneous network environment. In this way, more players can be served which can further increase the business potential of network games. The performance of the new method has been evaluated by using measurements and extensive simulations.

The utilization of the free capacity by the TCP congestion control is not perfect due to the complex fluctuations of the background traffic resulting in reduced throughput. Different frequency components of the background traffic have different effects on the TCP performance. In this paper we present a comprehensive TCP performance evaluation study to understand the nature of the TCP adaptivity mechanism in frequency domain. Through simulations we analyze the impact of network parameters and also some alternatives for compensation of the throughput reduction. Finally, based also on the analysis of measured TCP traffic we investigate the robustness of TCP to fluctuations occur in actual network traffic.