The paper focuses on dynamic resource provisioning in an all optical wide area Software Defined Network supporting energy-aware interplay between fog and cloud. The main contribution of this paper is a latencyAware policy for handling fog related traffic. The policy is combined with anycast strategies in order to form a complete approach able to handle fog and cloud traffic simultaneously. The proposed approach was compared with reference solutions with respect to latency of fog related requests, CO2emission and impact on network performance. It is shown that both latency and carbon footprint may be reduced without significant deterioration of network performance.

The paper focuses on cooperation between cloud and network operators, as well as on fitting particular routing strategies to various cloud services. Three cooperation models are presented, analyzed and compared in the paper: the proposed model and two widely used reference models. The main difference between the models is the set of information being exchanged between the involved parties. Additionally, we analyze the applicability of four fitting schemas for each considered model. It is shown that the proposed model, alongside with an appropriate fitting schema, is able to reduce the blocking probability of cloud services requests. At the same time, thanks to the use of green anycast strategies, it is able to significantly reduce carbon dioxide emission.

The paper focuses on dynamic resource provisioning which minimizes carbon footprint of data centers interconnected via optical networks. The main contribution of this paper is a schema of fitting energy aware anycast strategies to different types of cloud services in order to reduce greenhouse gases emission. The proposed schema was compared to the cases when all types of cloud services were handled using the same anycast strategy. It is shown that the proposed schema is able to significantly reduce greenhouse gases emission without significant deterioration of network performance.

The paper deals with the problem of optimizing the use of energy in a telecommunication network. We propose the power capping technology to effectively move power hungry tasks from nodes powered from non-renewable energy sources to nodes powered from renewable resources. The presented technology may also be used to optimize OPEX in case the cost of energy varies in different nodes. The presented results show, that the solution is extremely effective in meeting the aforementioned goals.

The paper focuses on dynamic resource provisioning which minimizes carbon footprint in optical networks. The main contribution of this paper consists of three anycast routing strategies aimed at reducing greenhouse gases emission by processing energy-intensive requests in data centers powered from renewable energy sources. The proposed strategies were compared with reference approaches with respect to CO2emission and their impact on network performance. It is shown that they are able to significantly reduce greenhouse gases emission and, simultaneously, improve network performance.

Measurement-based admission control routine in Flow-Aware Networks often leads to fair rate degradation, which negatively impacts the performance of ongoing flows. This problem can be solved with the limitation mechanism, however, only if properly configured. When the limitation mechanism is badly configured or the network conditions change rapidly the mechanism does not help, and sometimes, it can even exacerbate the problem. This paper shows that it is possible and more efficient to provide dynamic limitations, where the admission limit is not static, but dynamically calculated. It is shown that the proposed mechanism provides better results and is more robust. Also, the issue with configuration is diminished, since the mechanism automatically adapts to the current network congestion level.

This JSAC special issue is covering four emerging areas in communications: Area 1 - e-Health; Area 2 - Peer-to-Peer Networking; Area 3 - Vehicular Networks and Telematics Applications; and Area 4 - Social Networks. This is the first issue of a brand new series in JSAC focusing on "Emerging Technologies in Communications." These emerging technical areas were selected from ComSoc's Technical Committees and Technical Sub-Committees. (The second issue will be published later in 2013 on the following areas: Area 5 - Applications of Nanotechnologies in Communications; Area 6 - Autonomic Communications; Area 7 - Fiber Wireless Integration; and Area 8 - Nanoscale and Molecular Networking.) The papers selected are briefly summarized. For this issue, 55 regular papers were selected for publication out of 197 open-call submissions. Also included are three overview (invited) papers. It is devoted to reporting on cutting-edge research achievements covering the following four emerging technical areas in communications.

The paper addresses the problem of unwanted stabilization of a virtual topology in a dynamic multilayer network using the Packet-Packet-Lambda (PPL) routing strategy. The PPL strategy was already proposed in several papers and is perceived as a valuable solution to the multilayer traffic routing problem in a network with lightpaths created on demand. However, as it is shown, usage of the strategy may bring catastrophic results, since this strategy may create in some point of time an inefficient, extremely stable and long lasting logical topology, which results in a highly penalized network performance. Therefore, enhancements to the strategy are necessary. We show some directions for such improvements.

A new fairness concept for Flow-Aware Networks (FAN) is presented and analyzed in the paper. In contrast to basic FAN, where elastic flows are treated similarly, it assumes fair access to the resources for each user independently of how many flows it generates. A new method for estimating the values of the fair rate, which is the key congestion control parameter in FAN is also provided. The new solution allows for reducing the oscillations of the periodically measured values which influences more stable transmission in the network. The new method allows for describing the traffic in a more realistic way, simultaneously simplifying its implementation in the FAN routers. Finally, the results of the simulation analysis of FAN with the new fairness algorithm and the Remove and Accept Most Active Flows (RAMAF) congestion control mechanism are presented for two versions of the TCP protocol: NewReno and NewJersey.

Admission control routine in Flow-Aware Networks (FAN) may lead to severe fair rate degradation, which negatively impacts the performance of ongoing flows. To mitigate the problem, a predictive approach is proposed in this paper. Instead of waiting for the congestion to occur, proper admission control actions are performed when the system approaches the congestion state. After each network measurement, by analyzing the trend, the next values are predicted and actions are taken based on the predicted values. The proposed mechanism can be simultaneously used with the limitation mechanism providing a considerable performance increase.

Applicability of the concept of Flow-Aware Networking (FAN) for QoS assurance in wireless networks is studied in the paper. The link utilization efficiency for six versions of the TCP protocol in the FAN environment is investigated and compared. The second contribution of the paper is a new congestion control mechanism for FAN, referred to as RAMAF (Remove and Accept Most Active Flows), that fulfills the wireless link requirements. While in basic FAN, new flows cannot begin transmission in congestion, the new solution ensures short acceptance times of new streaming flows in a router regardless if the outgoing link is congested or not. As compared to other congestion control mechanisms proposed for FAN, RAMAF does not need to tune any parameters, and is thus automatic. Moreover, the proposed solution allows for more effective transmission of elastic flows in comparison to other approaches, and may be used in both wired and wireless FAN. The results of the carefully selected simulation experiments show that TCP NewJersey incorporated with the RAMAF mechanism meets the requirements of the wireless transmission in FAN.

A new proposal for Metropolitan Area Networks is presented and analyzed in the paper. The Flow-Aware Resilient Ring (FARR) combines the best features of two known network architectures: Flow-Aware Networking and Resilient Packet Ring. Transmitted traffic in FARR networks is based on implicitly classified flows which allows for providing Quality of Service guaranties in accordance with network neutrality rules. Moreover, the traffic with high priority is protected by the steering mechanism which ensures fast traffic redirection in time less than 50 ms. The advantages and weaknesses of the proposed architecture and new solutions to be used in it are described and analyzed in the paper. Moreover, it is shown that simultaneous implementation of all of them ensures fast, scalable and reliable transmission of streaming flows in FARR.

In this paper, it is shown that the admission control routine in Flow-Aware Networks (FAN) may lead to severe fair rate degradation, which negatively impacts the performance of streaming applications. In order to prevent this negative behavior, the limitation mechanism is proposed. The aim of the mechanism is to limit the maximum number of new flows that may be admitted on a link between any two consecutive network's automeasurements. The solution is efficient, viable and dramatically reduces the fair rate degradation.

A complete system which ensures reliable transmission of streaming flows in flow-aware networks (FAN) is presented and analyzed in the paper. A new congestion control mechanism, called RPAEF (remove and prioritize in access active elastic flows), is described in detail. It allows for fast acceptance of new streaming flows in the admission control (AC) block of FAN routers. The mechanism of limiting the number of new flows accepted in the AC block is provided as an answer to the scalability problem of all congestion control mechanisms proposed for FAN up to now. The global list of protected flows ensures that all streaming flows redirected from the primary route in case of a network element failure are immediately accepted in the first router on a backup route. The advantages and weaknesses of the proposed solutions are described and analyzed. Moreover, it is shown that the simultaneous implementation of all of them ensures fast, scalable and reliable transmission of streaming flows in FAN.

Due to the development of Next Generation Networks, leading to a multiservice transport layer with a multidomain environment, the importance of interconnection issues keeps growing. As the number of possible partners increases, operators face different routing options with regard to service quality and cost. We present the extension of the Least Cost Routing (LCR) concept which takes into account the reliability issues. As for a large number of variables the LCR problem is unsolvable, some heuristic algorithms to find the problem solution are proposed and evaluated. The solution helps to optimize connections among telecommunication operators by minimizing costs for served demands and protecting against network failures.

A new variation of the Flow-Aware Networking (FAN) concept is presented in the paper. The proposed solution is based on the Approximate Fair Dropping algorithm and called by us Approximate Flow-Aware Networking (AFAN). The simplicity of the new proposal is assured as a result of using only two FIFO queues for the scheduled packets. The impact of congestion control mechanisms proposed for Flow-Aware Networks on packet transmission in the overloaded network for the AFAN is also presented. The analysis shows how to choose the values of the parameters which decide of proper transmission characteristics of the network. The results of the research are compared with those obtained for the other two FAN architectures, with the PFQ (Priority Fair Queuing) and with the PDRR (Priority Deficit Round Robin) scheduling algorithms. The advantages and weaknesses of the new proposal are described and analyzed.

The impact of the control plane architecture on reliability of the GMPLS network is studied. To improve this reliability, a method based on the graph-theoretical dominating set problem is proposed. Several algorithms to select dominating nodes are presented and evaluated for four different network topologies by using simulation methods. It is shown that the service recovery time can be shortened by using the presented approach.

New concepts of the congestion control mechanisms in Flow-Aware Networks (FAN), which improve network performance and servicing of streaming flows, are proposed. The paper describes the congestion control mechanisms as an enhancement proposed for the Measurement Based Admission Control (MBAC) block in FAN. Three algorithms, i.e., RAEF (Remove Active Elastic Flows), RBAEF (Remove and Block Active Elastic Flows) and EFM (Enhanced Flushing Mechanism) focus on partial or total cleaning of the contents of Protected Flow List (PFL) in the congestion state. The mechanisms are discussed and performance evaluation results are presented.

The paper presents an analytical modular model of a DiffServ network supporting assured forwarding PHB. Individual DiffServ network elements such as meter/markers and droppers are modeled separately. Additional equations make it possible to describe behavior of a complex DiffServ network in mathematical terms. The modular model makes it possible to calculate several network performance parameters under various configurations and traffic volume. The model can be used for designing networks, is helpful in initial setting of configuration parameters and enables quick evaluation of network performance.

The paper is focused on an optical network in which the network operator needs to achieve differentiated connection blocking probability between different classes of lightpath requests. We propose and investigate several routing strategies, which may be used for that purpose, i.e., to maintain the blocking probability of the high priority requests below an assumed value in a broad range of network loads. Simulation results show that all proposed strategies achieve the stated goal, but their performance varies considerably. The paper presents differences in performance of individual strategies.

The paper proposes analytical models of two types of multi-RED queues, WRED and RIO-C, serving as droppers in DiffServ networks. Both are scrutinized under two types of configuration of threshold parameters: staggered and overlapped. The proposed analysis uses the M(n)/M/1/K and M(n)/D/1/K queuing models (queuing systems with balking). Additionally, in the case of RIO-C, a Bernoulli scheme is adopted in the model. The models are validated under two types of traffic: ftp-like and web-like. Recommendations on the selection of the most appropriate model are provided.

The 15 papers in this special issue focus on traffic engineering for multi-layer networks. The papers can be roughly divided into the following three groups: 1) New Approaches to Traffic Engineering, 2) Multi-layer Survivability, and 3) Specific Issues. Some of the specific issues include optimal topology design, multicast flow aggregation, load balancing, traffic grooming, and Ethernet-based technology.

Carriers conveying long-distance traffic face different routing options with regard to service quality and cost. Thus, least cost routing (LCR) solutions supporting optimal decisions become a very important issue in the inter-carrier operations. In this paper, we present this problem in the context of the long-distance traffic partner-carrier choice. The mathematical optimization problem is formulated. Especially, constraints (e.g., capacity, volume) related to the techniques used are considered. Different tariff formulations are taken into account as well.

In the paper we propose three route management strategies that allow an optical network operator to achieve differentiated service for two classes of lightpath requests. Simulation results show that all three strategies achieve the stated goal, which is to maintain blocking probability of high priority requests below an assumed value in a broad range of network loads. Moreover, the presented strategies require only little support from the control plane. Detailed algorithms are presented in two versions, one suitable for networks with centralized control, and the other for networks with distributed control.

A new concept Quality of Resilience (QoR) presented in this paper is based on the distinction in the reliability-related Quality of Service (QoS) parameters of the short-term quality factors and the long-term quality factors. The former parameters are called availability parameters and the latter are called QoR parameters. In one hand by dividing the service duration time into intervals, the service is considered available during a time interval, if the Service Level Agreement (SLA) between the user and the network operator is satisfied. In the other hand the long-term characteristics of the service are derived from the service downtime distribution. With the downtime histograms the asymptotic characteristics of the service can be represented both at the transport and service layers. Since the resilience mechanism implemented into the network match the transport layer downtime histograms, this new characterization of the QoS helps to measure the impacts of a given recovery scheme on the next generation services.

In the paper, theoretical formulas describing the reliability function as well as the mean time to failure for static p-cycles are derived. The practical conclusions concerning the design of p-cycles are drawn: p-cycles do not always have good properties from the reliability viewpoint; if an operator decides to use them, it has to ensure that the trade-off between sharing of resources and the availability is reasonable. An appropriate function on which such a choice can be based is introduced.

The paper presents mathematical models for meter/markers serving as DiffServ node elements. Models for two types of token bucket based meter/markers: single rate three color marker and two rate three color marker and a model for a rate estimator based meter/marker, i.e. time sliding window three color marker, are provided. Experimental validation of these models is also presented.

The paper presents a mathematical model of the TCP behavior in a DiffServ network and its empirical validation. The assured forwarding per hop behavior is modeled. The model is modular - individual components of DiffServ nodes are described by separate sets of equations. It makes it easy to develop a model of a system consisting of various types of DiffServ node elements (meters/markers, shaper/droppers) just by changing some of the equations. The input traffic is divided into several traffic aggregates each having a part of a bottleneck capacity dedicated to it.

Although the research on ATM elementary switches may be considered as close to the end, the problem of finding the optimum switching network still remains open. The main reason is that traditional switching networks (e.g., Clos switching networks) are not suitable for asynchronous environments, especially due to poor performance of last stages. Moreover, the load is not distributed equally over all switching elements inside networks of this class. In this paper a new type of switching network, referred to as polyhedral switching network is introduced. The key idea of a proposed solution is based on placing elementary switches on a sphere-shaped surface. Any set of adjacent switches forms a polygon, and the polygons, in turn, form a polyhedron. It is shown that such an arrangement is more convenient and suitable for ATM switching than switching networks known to date.

The paper presents a set of criteria which can be used for comparing, and in general analyzing, GDMO-based management information models (MIMs) designed for network configuration modeling. The criteria have been obtained by applying three selected MIMs, namely the NMF network model, the ITU-T M.3100 model and the ETSI GOM model, to a simple example SDH network. This process has highlighted similarities and differences between those MIMs which have been then summarized as comparison criteria. Those criteria can be used for assessing the difficulty of integrating management systems implementing different MIMs and hence having MIBs of different structure, for designing new MIMs and upgrading existing ones, and for choosing the most appropriate MIM for a given network and a given modeling case.

The paper describes various techniques to increase the speed of time-division switches to support rates of 600 Mb/s and beyond. The issue of multirate switching, essential to BISDN applications, is also addressed. Architectures of RAM- and shift-register-based time switches are discussed and compared. Some representative structures of time-division switch fabrics applicable to BISDN are presented.

There has been a great technological advancement in microelectronics and communications, which forces, among others, investigations on ATM switching architectures. Researchers keep on searching an efficient architecture, able to perform at a wide range of traffic loads and patterns, ensuring both minimum probability of cell loss as well as minimum cell delay. The authors propose one more architecture. The main goal is to create an architecture capable of operating at very high data rates. As the result, the switch uses the fastest method of buffer management, and is characterized by sufficiently low probability of blocking and short delay time. Appropriate simulation results are presented and some implementation issues are discussed.

A growable ATM or fast packet switching fabric architecture is proposed. The fabric has a two-sided structure and allows for the modular growth of its size from small to very large number of ports. The fabric is expanded by connecting additional basic fabrics to the existing structure by some interconnecting modules. The expansion process does not disturb the existing cabling. The architecture makes it also possible to adjust the system throughput. The hardware complexity of the new fabric is moderate and is discussed in the paper. Some important implementation issues such as the physical design and reliability are also presented in the paper.>

A digital telecommunication network, TELBANK, developed and operated by a consortium of Polish banks is described. Four TELBANK subnetworks, i.e., TELBANK-M, TELBANK-T, TELBANK-P, and TELBANK-VSAT are presented. TELBANK-M is a subnetwork offering long-distance transmission channels. TELBANK-P and TELBANK-T are packet-switched and circuit-switched subnetworks, respectively. TELBANK-VSAT is a satellite communication network. Three examples of TELBANK use for connections to external services are also presented.>

A class of photonic switching networks composed of directional couplers is proposed. These networks require only two switching stages containing active elements. Both point-to-point and multicast architectures are presented. Various characteristics of the networks are compared to those of networks previously known with respect to insertion loss, SNR, number of crossovers, and number of active elements.>

The following classes of connecting networks, based on their combinatorial properties, have been previously defined: networks nonblocking in the strict sense, networks nonblocking in the wide sense, rearrangeable networks, and blocking networks. To these the authors add the class of repackable networks, i.e., networks in which blocking can be avoided by using call repacking control algorithms. The conditions under which a three-stage Clos network is repackable are formulated and proved. The numbers of middle-stage switches in all network classes are compared, and the differences between repackable and rearrangeable networks are discussed.>

Methods of fault detection and location for separate digital switching matrices as well as a wide class of switching networks are proposed. First, it is assumed that stuck-type faults can affect only a single component of a network. A model of a switching matrix is presented and fault equivalences are shown. After the discussion of single-switch diagnosis, test vectors and connection settings for identification of a single faulty switch in multistage networks are proposed. To avoid the masking effects, the notions of an H-test vector and an H-assignment are introduced. In the proposed approach, tests for space and time switching are performed separately. The problem of multiple fault detection is also considered.>

In this paper, the problem of selecting minimum cost switching networks composed of digital switching matrices is discussed. It is shown that the conventional continuous optimization method may lead to nonoptimal structures. The optimization problem is solved by using the dynamic programming principle. The relevant formulas are derived and samplings of the results of optimization are presented. Different network structures are compared and discussed.

One-sided switching networks composed of uniform digital switching matrices are considered. Such matrices mix in the same integrated circuit or in the same printed circuit board time and space switching. The conditions under which one-sided networks are nonblocking and rearrangeable are discussed.

In the paper, four theorems describing conditions under which a three-stage multiconnection network is nonblocking for different routing strategies are formulated and proved. These theorems hold for any values of the structural parameters of a network. For some networks having a small number of outer stage switches, these theorems give significantly lower numbers of required middle stage switches than theorems currently known.

A simple algorithm for the control of rearrangeable switching networks\nu(n, n, r)is given. This algorithm makes it possible to compute the switch settings that implement a given Permutation. The algorithm takes serial timeO(nr^{2})and consists of one phase, i.e., it does not require iterations. The version of the algorithm which can be used for group switching is also presented.

Significant advances in semiconductor technology make it possible to construct new components called digital symmetrical matrices, which mix in the same integrated circuit, time and space switching. In this paper, structures of multistage nonblocking networks composed of such matrices are described. The condition under which the network will be nonblocking is formulated, and the relationship between the maximal capacity of the network and the number of switching stages is discussed. Formulas for designing optimal threeand five-stage networks are derived. It is shown that general multistage networks should have as few stages as possible for minimizing the cost.

Known rearrangement algorithms proposed by Slepian and Paull (method 1) involve two middle switches of the three-stage Clos network. These switches are chosen arbitrarily. However, the middle switches selection rules affect the rearrangement process. In this paper four methods for choosing the switches were studied. Simulation results and analysis have shown that the allocation of the least used switches can decrease the volume of computation required for rearrangements. In the case of limited rearrangement, this method improves the blocking performance of a network.

The class of switching networks, composed of stages of identical square switches arranged symmetrically around a center stage (Benes' networks), is considered. The synthesis problem of selecting from this class a rearrangeable network with the minimum number of crosspoints is solved. Benes' approach, (using relation of covering), is taken into consideration. The new structures of the networks are in some cases different than those already known. Modifications of Benes' theorems result in lowering the number of stages of a switching network with the minimum number of crosspoints. This lowering can bear some practical significance, which could simplify the control devices. A model comparison is given of the known and new results.