Opaque virtual network mapping algorithms based on available spectrum adjacency for elastic optical networks

Optical network virtualization enables multiple virtual optical networks constructed for different infrastructure users(renters) or applications to coexist over a physical infrastructure. Virtual optical network(VON) mapping algorithm is used to allocate necessary resources in the physical infrastructure to the VON requests(VRs). In this paper, we investigate the opaque VON mapping problems in elastic optical networks(EONs). Based on the concept of available spectrum adjacency(AvS A) on links or paths, we consider both node resource and AvS A on links for node mapping, and present a link mapping method which chooses the routing and spectrum plan whose AvS A on paths is the largest among all the candidates. Finally, the overall VON mapping algorithm(i.e., AvS A-opaque VON mapping, AvS A-OVONM) coordinated node and link mapping is proposed. Extensive simulations are conducted and the results show that AvS A-OVONM has better performance of blocking probability and revenue-to-cost ratio than current algorithms.     

WDM光网络动态虚拟拓扑重构算法

针对波分复用光纤网络上业务流量动态改变的问题,为了使光纤网络能支持更多的业务连接,需要对虚拟拓扑进行重构。基于链路最大负载和包平均跳步距离,利用混合线性规划公式对重构问题进行描述,在此基础上提出一个自适应拓扑重构算法,达到提高网络吞吐量的目的。仿真结果表明,该算法可以有效地改善网络性能。     

An elastic compensation model for frame-based scheduling algorithms in wireless networks

Scheduling algorithms for high-speed wireless networks need to be simple to implement for serving packets while ensuring quality-of-service (QoS). The ordinary frame-based scheduling principle is well-known for providing fair service with low implementation complexity. However, existing frame-based scheduling algorithms cannot properly handle location-dependent burst errors in wireless networks. To utilize the advantages of frame-based scheduling algorithms in error-prone wireless networks, we propose an elastic compensation model that provides not only smooth compensations without any service disruptions of flows but also flexible compensations to flows that experience frequent errors to provide flows with fairness of service. From our analysis and simulation studies, we found that the proposed compensation model shows smooth compensation performance without any service disruption periods and good fairness performance when channel errors occur.     

Routing wavelength and time-slot reassignment algorithms for TDM based optical WDM networks

In this paper, we consider the problem of maximizing the time of first lightpath request rejection, T in the circuit-switched time division multiplexed (TDM) wavelength-routed (WR) optical WDM networks. TDM is incorporated into WDM, to increase the channel utilization when the carried traffic does not require the entire channel bandwidth. In TDM–WDM network, multiple sessions are multiplexed on each wavelength by assigning a sub-set of the TDM slots to each session. Thus, given a session request with a specified bandwidth, a lightpath has to be established by using the routing, wavelength and time-slot assignment (RWTA) algorithms. If the lightpath cannot be established, lightpath request rejection or call blocking occurs. As each lightpath is substantial revenue and long-lived, lightpath request rejection is highly unfavourable in the optical backbone networks. In this paper, we are proposing an intelligent routing, wavelength and time-slot reassignment algorithm for multi-rate traffic demands, where, when a call gets blocked, the already established calls in the network are rerouted, wavelength and time-slot reassigned so as to accommodate the blocked call. Since we are talking of slow arrivals and long holding times for the lightpaths, it is possible to do this reassignment while provisioning a new call. Simulation based analyses are used to study the performance of the proposed reassignment algorithm. The results show that the proposed reassignment algorithm can be used to maximize the time of first call blocking, thereby accommodating more calls in the network before upgrading the network capacity.     

On providing elastic QoS in optical burst switched networks

Integrated services schemes have not been used to support QoS in OBS networks, largely because of the high control overhead involved in monitoring bursts continuously at intermediate nodes. In this paper, we propose for the first time, an integrated services scheme to support elastic QoS in OBS networks. Our scheme relieves intermediate nodes of the burden of monitoring individual bursts continuously, thus largely decreasing the control overhead. We show how absolute bandwidth guarantees can be provided, and how bandwidth used by each connection can be made elastic. We demonstrate the effectiveness of our scheme using both analytical and simulation studies.     

Social networks and genetic algorithms to choose committees with independent members

Choosing committees with independent members in social networks can be regarded as a group selection problem where independence, as the main selection criterion, can be measured by the social distance between group members. Although there are many solutions for the group selection problem in social networks, such as target set selection or community detection, none of them have proposed an approach to select committee members based on independence as group performance measure. In this work, we propose a novel approach for independent node group selection in social networks. This approach defines an independence group function and a genetic algorithm in order to optimize it. We present a case study where we build a real social network with on-line available data extracted from a Research and Development (R&D) public agency, and then we compare selected groups with existing committees of the same agency. Results show that the proposed approach can generate committees that improve group independence compared with existing committees.     

Parallel algorithms for tensor completion in the CP format

Low-rank tensor completion addresses the task of filling in missing entries in multi-dimensional data. It has proven its versatility in numerous applications, including context-aware recommender systems and multivariate function learning. To handle large-scale datasets and applications that feature high dimensions, the development of distributed algorithms is central. In this work, we propose novel, highly scalable algorithms based on a combination of the canonical polyadic (CP) tensor format with block coordinate descent methods. Although similar algorithms have been proposed for the matrix case, the case of higher dimensions gives rise to a number of new challenges and requires a different paradigm for data distribution. The convergence of our algorithms is analyzed and numerical experiments illustrate their performance on distributed-memory architectures for tensors from a range of different applications.     

Impairment-aware fixed-alternate BSR routing heuristics applied to elastic optical networks

The Journal of Supercomputing - The introduction of elastic optical networks (EONs) leads to new challenges, such as the problem of routing, modulation level, and spectrum assignment. In...     

A subcarrier-slot partition scheme with first-last fit spectrum allocation for elastic optical networks

Bandwidth fragmentation refers to the existence of non-aligned and non-contiguous available subcarrier slots in elastic optical networks. Since spectrum for a connection must be allocated to contiguous slots and aligned along the routing path, non-contiguous and non-aligned available slots could cause call blocking. This paper proposes a subcarrier-slot partition scheme with first-last fit spectrum allocation for elastic optical networks in order to increase the number of contiguous aligned available slots, and hence the blocking probability in the network is suppressed. The partition approach creates more aligned available slots by separating the spectrum allocation of the non-disjoint connections. The first-last fit allocation policy creates more contiguous aligned available slots between two partitions. Simulation results show that the partition approach with first-last fit allocation policy provides lower blocking probability than that of the non-partition approach and the partition approach with other spectrum allocation policies. Furthermore, simulation results indicate that the blocking probability in the network using the proposed scheme improves as the number of partitions is reduced.     

Nonholonomic orthogonal learning algorithms for blind source separation

Independent component analysis or blind source separation extracts independent signals from their linear mixtures without assuming prior knowledge of their mixing coefficients. It is known that the independent signals in the observed mixtures can be successfully extracted except for their order and scales. In order to resolve the indeterminacy of scales, most learning algorithms impose some constraints on the magnitudes of the recovered signals. However, when the source signals are nonstationary and their average magnitudes change rapidly, the constraints force a rapid change in the magnitude of the separating matrix. This is the case with most applications (e.g., speech sounds, electroencephalogram signals). It is known that this causes numerical instability in some cases. In order to resolve this difficulty, this article introduces new nonholonomic constraints in the learning algorithm. This is motivated by the geometrical consideration that the directions of change in the separating matrix should be orthogonal to the equivalence class of separating matrices due to the scaling indeterminacy. These constraints are proved to be nonholonomic, so that the proposed algorithm is able to adapt to rapid or intermittent changes in the magnitudes of the source signals. The proposed algorithm works well even when the number of the sources is overestimated, whereas the existent algorithms do not (assuming the sensor noise is negligibly small), because they amplify the null components not included in the sources. Computer simulations confirm this desirable property.     

Randomized self-routing algorithms for Clos networks

As the VLSI technology makes large crossbar switches affordable, Clos networks have become a feasible option of large interconnection networks. However, to make these networks practical and useful, efficient routing algorithms need to be developed. This paper will develop and study several randomized routing algorithms for Clos networks. The algorithms are based on the idea that if the first column of Clos is set to some configuration somehow, then the resulting network becomes self-routed using the destination addresses. Each of the randomized algorithms sets the first column to a configuration selected by a random process. The algorithms are then self-routed and take no computation time to set the switches. Probabilistic analysis and simulation measurements of the communication delay of permutation routing are conducted. It is shown that the communication delay of any permutation is 3–6 cycles in networks of up to 1024 processors. Although other routing algorithms route arbitrary permutations in one cycle over Clos/Benes networks and 2 cycles over δ networks, these algorithms take prohibitively large times to compute the appropriate switch settings, while our randomized algorithms are self-routed and spend NO time on computing the switch settings. This makes our algorithms superior to any universal nonrandomized routing algorithm for Clos/Benes networks or δ networks. The speed, universality and ease of implementation of our randomized algorithms make Clos networks highly attractive for large parallel computer systems.     

QoS-based routing algorithms for ATM networks

This paper presents a planned routing algorithm (PRA) and a hierarchical routing algorithm (HRA) for ATM networks. The PRA can establish the multicast tree with the presence of bandwidth and delay constraints. The HRA can be compliant with the PNNI specification from the ATM Forum. It uses an adaptive and iterative path search approach and takes advantage of the PNNI hierarchical network structure to reduce path computation complexity and maximize network throughput. The performances of the PRA and HRA are evaluated by simulations. The simulation results show that the PRA can provide the best performance while the complexity is acceptable and the HRA can reduce processing time and improve network utilization, and both are suited for QoS requirements of ATM networks’ routing.     

Multi-period traffic on elastic optical networks planning: alleviating the capacity crunch

Advances in optical data transmission technology have allowed the current expansion of bandwidth-demanding services over the Internet. Also, the emergence of orthogonal frequency-division multiplexing (OFDM) has opened the possibility of increasing the network spectral efficiency by solving the routing, modulation and spectrum assignment (RMSA) problem. Recently, investigators have examined the effects of multiple demands or multiple virtual topologies when they are requested at different time periods over a single physical substrate. That makes the RMSA harder and with many more instances. Such analysis is required because network traffic does not remain static along time, and the demand can increase considerably as new user services arise. Therefore, planning the network considering a multi-period study becomes essential, since it can prevent a case where demands may exceed the bandwidth capacity and cause request blocking in future periods. In this work, we provide a novel mixed integer linear programming (MILP) formulation to solve the RMSA problem for several t periods of demands. This model can be used not only to find the solutions to minimize the used capacity, but also as an efficient method of network planning, since it can estimate with a single formulation and a single iteration the point of resource exhaustion in each period t. The results are found for a small network, and they show the efficiency of the proposed MILP formulation. We also propose an alternative version of this formulation with predefined paths, which is less computationally demanding. The results of this study are compared to a step-by-step planning, where the strategy is a decomposition method that breaks the previous formulation into t steps. Comparing the results of the two strategies, it can be seen that the single multi-period formulation is a good strategy to solve the problem. By contrast, the step-by-step strategy may require reconfigurations and eventual interruptions in the network, from a step to another one.

     

后非线性混叠信号盲源分离算法综述

结合非线性盲源分离研究不断发展的现状,选取最常见的后非线性混叠信号盲源分离问题为对象,综述其算法,介绍了解混叠模型,说明了基于广义Gram-Schmit正交化构造解的存在性及非唯一性.在此基础上,阐释了分离方法和思路,概述了基于互信息最小化的独立性测度,并分析评述了不断涌现的后非线性盲源分离典型算法.最后指出,目前关于后非线性盲源分离算法的研究存在的共性问题,并对进一步的研究方向进行了展望.     

Fast and robust fixed-point algorithms for independent componentanalysis

Independent component analysis (ICA) is a statistical method for transforming an observed multidimensional random vector into components that are statistically as independent from each other as possible. We use a combination of two different approaches for linear ICA: Comon's information theoretic approach and the projection pursuit approach. Using maximum entropy approximations of differential entropy, we introduce a family of new contrast functions for ICA. These contrast functions enable both the estimation of the whole decomposition by minimizing mutual information, and estimation of individual independent components as projection pursuit directions. The statistical properties of the estimators based on such contrast functions are analyzed under the assumption of the linear mixture model, and it is shown how to choose contrast functions that are robust and/or of minimum variance. Finally, we introduce simple fixed-point algorithms for practical optimization of the contrast functions     

Diffusion learning algorithms for feedforward neural networks

The problem of training feedforward neural networks is considered. To solve it, new algorithms are proposed. They are based on the asymptotic analysis of the extended Kalman filter (EKF) and on a separable network structure. Linear weights are interpreted as diffusion random variables with zero expectation and a covariance matrix proportional to an arbitrarily large parameter λ. Asymptotic expressions for the EKF are derived as λ→∞. They are called diffusion learning algorithms (DLAs). It is shown that they are robust with respect to the accumulation of rounding errors in contrast to their prototype EKF with a large but finite λ and that, under certain simplifying assumptions, an extreme learning machine (ELM) algorithm can be obtained from a DLA. A numerical example shows that the accuracy of a DLA may be higher than that of an ELM algorithm.     

New models and algorithms for programmable networks

In todays IP networks most of the network control and management tasks are performed at the end points. As a result, many important network functions cannot be optimized due to lack of sufficient support from the network. The growing need for quality guaranteed services brought on suggestions to add more computational power to the network elements.This paper studies the algorithmic power of networks whose routers are capable of performing complex tasks. It presents a new model that captures the hop-by-hop datagram forwarding mechanism deployed in todays IP networks, as well as the ability to perform complex computations in network elements as proposed in the active networks paradigm. Using our framework, we present and analyze distributed algorithms for basic problems that arise in the control and management of IP networks. These problems include: route discovery, message dissemination, multicast, topology discovery, and bottleneck detection.Our results prove that, although adding computation power to the routers increases the message delay, it shortens the completion time for many tasks. The suggested model can be used to evaluate the contribution of added features to a router, and allows the formal comparison of different proposed architectures.     

On parallel scan-conversion algorithms for transputer networks

Two fundamental scan line procedures are considered from the point of view of parallel implementation on transputer networks. The paper describes algorithms for the scan conversion of polygons, and the related hidden surface elimination problem for three-dimensional models with polygonal data structures. In each case parallel designs have been implemented and timed in various configurations against a functionally equivalent, but not necessarily optimized, single processor code. The results presented demonstrate that scan line algorithms can be efficiently implemented on suitably configured networks of transputers.     

Some simple distributed algorithms for sparse networks

Summary. We give simple, deterministic, distributed algorithms for computing maximal matchings, maximal independent sets and colourings. We show that edge colourings with at most colours, and maximal matchings can be computed within deterministic rounds, where is the maximum degree of the network. We also show how to find maximal independent sets and -vertex colourings within deterministic rounds. All hidden constants are very small and the algorithms are very simple. Received: August 2000 / Accepted: November 2000