Frontiers in Internet Technologies

Server-centric data center architecture has been proposed to provide high throughput, scalable construction and error tolerance with commodity servers and switches for cloud data centers. To fully utilize those advantages of server-centric data center, an effective routing algorithm to find high quality multiple paths in Server-centric network is needed. However, current routing algorithms cannot achieve this completely: (1) the state-of-art routing algorithms in server-centric data center just consider hop count when selecting paths; (2) traditional multi-constraint QoS routing algorithms only find one feasible path and are usually switch-oriented; (3) present multi-path algorithms cannot guarantee the performance of the founded paths. In this paper, we propose a multi-constrained routing algorithm for server-centric data centers, named Server-Centric Multi-Constrained Routing Algorithm (SCRAT). This algorithm exploits the topology features of the Server-Centric data center to decrease the algorithm complexity and returns optimal and feasible paths simultaneously. In simulations, SCRAT has a very high probability (more than 96 %) to find the exact optimal path, and the cost of the optimal path found in SCRAT is about 10 % less compared with path found in previous TS_MCOP. Compared with previous MPTCP, SCRAT reduces the path delay by 18 % less and increase the bandwidth by 20 %.

Network coding has gained wide attention nowadays for its significant advantages on many aspects compared with traditional routing mechanism. However, if there are malicious nodes launching pollution attacks by tampering or forging data packets in the communication network, the sink nodes will suffer from failure decoding, together with serious results such as bandwidth wasting, longer transmission delay and increasing computation overheads. The original null keys based pollution detection scheme cannot efficiently defend against pollution attacks when the system has colluding attackers because of high communication overheads. Therefore, we firstly define the concept of complete null space, with the property that no pollution packets can pass its verification. We then propose the idea of partial position detection and design an algorithm to construct short null keys. Secondly, we provide a short null keys based pollution detection scheme with network coding, which has lower overheads compared with the original null keys based pollution detection scheme in composing complete null space. Finally, rigorous theoretical proofs are given to analyze the security of the designed scheme.

Traditional data centers, which are based on wired networks, have been installed many years. However, they often suffer from high cost of construction and management. The data-intensive workloads motivate modern data centers to utilize higher-bandwidth networks like 10 Gb Ethernet (10 GbE). Traditional wired data centers, suffering from complex wiring and management, can not always satisfy this requirement. Wireless data centers have some advantages such as economical installing and no worry about wiring. This paper proposes a new completely wireless data center, which has good performance and high fault tolerance to the network failures. We propose a novel data center design including intra-rack and inter-rack architectures. Especially, inter-rack data transmission can be finished by one hop when we transmit data via the ceiling. Data center, built with this methodology, is significantly convenient to be deployed and managed, which avoids the complex wiring and too much material cost. Extensive simulations have been conducted to verify the effectiveness of our proposal. Our novel design of wireless data center outperforms Cayley data center on bandwidth and good fault tolerance as well as reducing hotspot in the network.

Privacy protection has become one of the important issues for location-based services (LBS) nowadays. In order to meet the requirements of humanization, security and quick response, this paper proposes an improved personalized k-anonymous location privacy protection algorithm with fake position generation mechanism. Compared to the normal personalized k-anonymity algorithm, our improved algorithm has higher success rate of anonymity. By generating fake queries for the source queries that expire, our algorithm guarantees that no source query will be dropped, namely all the source queries can get anonymized. The experimental results show that the algorithm proposed by this paper is able to achieve better performance in terms of success rate of anonymity.

Node localization is important in many WSN applications. Most of existing algorithms are not applicable for networks in a concave area. To address this problem, a novel localization algorithm is proposed. The key of this algorithm is the design of neighborhood function which only uses distances between neighbor nodes to generate a new set of estimated locations from an old set. Both range-based and range-free localization can use it to estimate locations of nodes if distances between neighbor nodes are measured by hardware or estimated by distance estimation algorithms. Simulation results indicate that this algorithm can achieve accurate and reliable localization results in a concave area.

Traditional networks make routing decisions based only on the destination address. This dramatically limits the number of ser-vices that a network can provide. However, in some of them, the demand for more flexible routing protocols has increased. For example, home and enterprise networks, as well as datacenters, need to support multi-homing and/or role-based access control. This is where two-dimensional forwarding tables step in. In this approach, routing decisions are based on the destination and the source addresses. However, this will increase the lookup time and make the memory use skyrocket if implemented carelessly. In this paper, we evaluate and compare different designs to find the best way to tackle this problem.

A number of new Clos data center networking fabrics have been proposed recently by using commodity off-the-shelf hardware. In Clos networks, randomized routing is typically used to achieve load balance. However, with typical communication patterns, pure randomized routing design cannot fully exploit bandwidth provided by Clos topologies. This paper strives to evaluate the performance of both randomized routing and its alternatives. As an incremental improvement, periodic path renewal is discussed first. We then discuss adaptive routing, together with its implementation consideration and costs. To reduce overhead of pure adaptive routing, a mixed routing scheme is also discussed. Through detailed discussions and comprehensive evaluations, we provide designers with useful insight and a range of options for configuring the routing scheme.

Because of dynamic nature of the application layer multicast node in mobile ALM, the node easily loss. The loss of upper-middle-class multicast tree nodes has a wide range of influence, but the loss of lower nodes, especially the leaf node has a small range of influence. According to this characteristic, Hierarchical Adaptive Recovery (HAR) algorithm is proposed. Firstly HAR stipulates the judging method of core nodes and primary nodes, namely hierarchical method; when the node is missing, based on different nodes loss, it can adopt different recovery strategy to renew the multicast tree. Also it can make the multicast tree more stable and get higher node recovery efficiency. Through the experimental comparison, it is proved that HAR has lower rejoining delay.

With the emergence of Distributed Router, a novel multicast forwarding method called NSFA (Novel Storage and Forwarding Approach) is proposed as the original forwarding approach TSFA (Traditional Storage and Forwarding Approach) has been out of date. In NSFA a new forwarding table model is developed, and what determines the times that packets are switched in the Internal High Speed Switched Networks is the number of outgoing line cards instead of the number of outgoing interfaces. In NSFA, compared with the previous method, the total number of times that packets are switched is considerable lower, which is validated in experiment. Through comparing NSFA and TSFA, it can be found that the memory consumption of NSFA is considerably less than TSFA.

As peer-to-peer (P2P) technology booms lots of problems arise such as rampant piracy, congestion, low quality etc. Thus, accurate identification of P2P traffic makes great sense for efficient network management. As one of the optimal classifiers, support vector machine (SVM) has been successfully used in P2P traffic identification. However, the performance of SVM is largely dependent on its parameters and the traditional tuning methods are inefficient. In the paper, a novel hybrid method to optimize parameters of SVM based on cuckoo search algorithm combined with particle swarm optimization algorithm is proposed. The first stage of the proposed approach is to tune the best parameters for SVM with training data. Subsequently, the SVM configured with the best parameters is employed to identify P2P traffic. In the end, we demonstrate the effectiveness of our approach on-campus traffic traces. Experimental results indicate that the proposed method outperforms SVM based on genetic algorithm, particle swarm optimization algorithm and cuckoo search algorithm.