Search

A connection between open and closed many-server queueing systems is examined. Two limits are considered: (i) the number and reliability of machines (customers) increase simultaneously while the offered load remains constant (Poisson limit), and (ii) the number of machines (customers) and repairmen (servers) increase while the utilization remains close to unity (QED limit). It is argued that the two limits are interchangeable.

In the modern data networks, the demand is increasing for interruption free, high speed, and best quality of services to perform successful communications. This paper emphasizes on the high priority (HP) traffic like video conferencing, video streaming, telemedicine, teleconferencing, and interactive gaming for the real time communications. To achieve the best quality data communication for the HP traffic, in this paper, we propose a new dynamic bandwidth allocation (DBA) algorithm called dynamic hybrid slot-size bandwidth allocation (DHSSBA) algorithm, which is the modified version of existing hybrid slot-size/rate (HSSR) algorithm. The performance of the proposed scheme has been analyzed by using numerical simulations in terms of the end-to-end packet delay, absolute delay variation, jitter, and throughput for both the HP and best effort (BE) traffic. We have also compared the simulation results of the proposed scheme with those of the existing modified HSSR (MHSSR), delay variation guaranteed protocol (DVGP), and HSSR schemes. From the comparison of the simulation results, at the highest offered load of 1.0, it can be seen that for the HP traffic the proposed scheme provides 60%, 50%, and 20% less average end-to-end delay than the existing HSSR, DVGP, and MHSSR schemes, respectively. At the highest offered load of 1.0, the proposed scheme also provides about 35% higher throughput than both the HSSR and DVGP schemes with 45% and 35% less jitter variation than the HSSR and DVGP schemes, respectively.

It is an effective approach for comprehending network performance is to develop a mathematical model because complex relationship between system parameters and performance can be obtained explicitly. This paper presents generalized analytical expressions for end-to-end throughput of IEEE 802.11 string-topology multi-hop networks. For obtaining expressions, a relationship between the durations of the backoff-timer (BT) decrements and frame transmission is expressed by integrating modified Bianchi’s Markov-chain model and airtime expression. Additionally, the buffer queueing of each node is expressed by applying the queueing theory. The analytical expressions obtained in this paper provide end-to-end throughput for any hop number, any frame length, and any offered load, including most of analytical expressions presented in previous papers. The analytical results agree with simulation results quantitatively, which shows the verifications of the analytical expressions.

We consider a batch arrival M X ∕ G ∕ 1 queue with impatient customers. The loss probability is expressed in terms of the stationary waiting time distribution for the standard M X ∕ G ∕ 1 queue with no impatience. But this expression is only applicable when the offered load ρ is less than 1. We give a formula for the loss probability applicable for any values of ρ > 0 , by proving that the loss probability is analytic in ρ on ( 0 , ∞ ) through a Girsanov-type change of measure.

The massive Internet of Things (IoT) scenario refers to a huge number of Machine Type Communications (MTC) characterized by sporadic transmissions of small-sized packets. In order to manage the uplink radio resource allocation in cellular networks, several access schemes have been proposed. These schemes are mainly based on a grant-based Random Access (RA) procedure and proper load-aware access controls, e.g., Access Class Barring (ACB) techniques, dynamic uplink radio resource allocation, and so on. The development of an efficient approach to estimate the traffic load is extremely important for the proper functioning of these access schemes. With the ever-increasing number of transmitting MTC devices, expected with Beyond 5G (B5G) and 6G networks, additional challenges to obtain a correct real-time traffic load estimation are posed. Deep learning techniques, in this context, offer learning ability and optimization capability to properly support this scenario. In this paper, we propose a current access attempts estimation, based on Deep Neural Network (DNN), which accepts as input only the information really available at the next generation Node B (gNB). The network was trained and tested with a dataset properly created and composed by more than 21 million points. The DNN-based traffic load estimation method is then compared with other benchmark schemes available in the literature, in terms of regression accuracy, both in a static analysis, which considers a stand-alone RA cycle, and through a long-term analysis with a time-varying offered load. The latter analysis was performed both by adopting a theoretical arrival process proposed by 3GPP, and by using traces of real traffic data.

Long queues in emergency departments (EDs) lead to overcrowding, a phenomenon that can potentially compromise patient care when medical interventions are delayed. There are several causes of this problem, one of which is inadequate resource allocation. In this paper, we propose using a modified version of the square root staffing (SRS) rule to satisfy the probability of delay target. We use the concepts of kinetics and biological modeling to approximate the fluid behavior of the queueing process. We are then able to estimate the offered load and the appropriate service grade necessary to construct a staffing policy that meets the target. Additionally, we show how to utilize Pontryagin’s maximum principle to find the optimal number of providers that minimizes delay and staffing costs. Finally, we demonstrate the implementation of our model using data from a hospital in upstate New York.

This paper proposes a transceiver system for automatic tracking and dynamic routing for free space optical (FSO) communication. The proposed transceiver architecture has M transmitters and M receivers in a M×M configuration that has the capability to dynamically orient themselves at different angles to establish a link between two nodes. It also addresses the problem of optimizing and designing logical topologies for FSO communications using tracking transceivers. Whereas the physical topology consists of nodes, transmitters and receivers that can orient themselves in different directions, the logical topology consists of a set of lightpaths and the set of nodes originating or terminating the lightpaths. A lightpath in a logical topology is a combination of one or more physical links using transmitters and receivers with the same wavelength. For a given physical topology and traffic pattern, the main challenge is to establish an optimal logical topology so as to minimize congestion. We formulate a mixed integer linear programming (MILP) methodology for constructing logical topologies in FSO networks with tracking transceivers. The main objective of the mathematical formulation is to minimize the maximum offered load on any physical link so as to minimize congestion in the network. Given a fixed number of resources, in comparison to a fixed FSO architecture, the proposed architecture reduces the blocking probability further as well as increases the percentage recovery of traffic affected by a single link failure.

Problem Definition: We determine the jointly optimal service capacity and retrial intervals between unsuccessful service attempts for a major provider of cloud computing services. Allocating sufficient capacity to cloud services is a challenging task because demand is time-varying. Thus, most firms have been expanding their capacity with little regard to the consequences associated with idle resources, such as excessive energy consumption and excess costs. Academic/Practical Relevance: We model the system as a multi-station queueing network where the arrival rate of jobs is time-varying and the servers represent CPU cores. Jobs are infinitely impatient and those that are not immediately serviced may retry several times before permanently abandoning the system. We introduce an offered load approximation that allows us to construct a recursive representation of the offered load function which describes the fluid dynamics of the system. Methodology: We develop a calculus-of-variation approach to minimize the total functional variation of the constructed offered load function. We show that an optimal policy can be efficiently obtained and prove that it is similar to maximizing the penalized system throughput. Results: Using a data set of cloud computing requests over a representative 24-hour period from a typical service of our partner organization, we show that our optimal policy results in a 10% reduction in capacity. We also demonstrate that small changes to their service-level agreements may elicit additional savings. Managerial Implications: Our model can help reduce idle capacity and has implications for managing more sustainable and environmentally friendly cloud computing services. It may also help to explain why so much global cloud capacity is typically idle. That is, in order to satisfy service level agreements encouraging retrial jobs to be processed during off-peak periods while also ensuring that they have short wait times, providers must provision large amounts of capacity.

Recently, high port-count optical packet switches, based on the Arrayed Waveguide Grating (AWG) device, have been proposed for providing the very high bandwidth and connectivity degree that will be required in future datacentre interconnection networks. A practical, cost-effective realisation of such a switch requires that over-subscription of the AWG cross-connect output ports be allowed, with the resulting packet contention being resolved using a packet buffering facility in the switch. In this paper, we propose an efficient AWG switch architecture of this type that uses a recirculating optical packet buffering scheme and we develop analytic modelling methods to estimate the switch packet blocking probability and packet buffering delay. The analytic model resolves internal traffic flows in the switch as Markov-modulated Poisson Processes (MMPP) and also allows modelling of offered load to the switch as a MMPP process. Taking this model of the switch, we first show its accuracy compared to discrete event simulations and then use it to develop a heuristic optimisation algorithm to dimension the switch hardware components, with the objective of minimising the operating cost by optimally dimensioning the main power consuming components, subject to constraints on the maximum contention probability and average and maximum propagation delays. The resulting optimised switch design demonstrates the proposed architecture to be a promising candidate for future high-port count datacentre switches.

This study aims to put forward an extensive discussion about the increasing demand for available bandwidth to serve the multiple types of traffic in modern wavelength division multiplexing (WDM) metropolitan area networks (MANs). A traffic-depended multi-buffer node architecture in conjunction with an efficient asynchronous transmission WDM access (WDMA) protocol to serve the variable size Internet packets in ring MANs is proposed. The structure of the multi-buffer node architecture is determined by the probability distribution of each packet size category in the MAN traffic, providing storage and dropping events equity among the nodes. The adopted WDMA algorithm satisfies the requirement for high performance efficiency especially under high offered load, by taking care to optimally face the bandwidth fragmentation problem and to maximize the bandwidth exploitation, while it effectively avoids both the packets collisions over the wavelengths and the destination conflicts. Numerical results prove that the proposed network model achieves throughput improvement up to 334% as compared with the relative study of Pranggono and Elmirghani (2011). An analytical framework is developed for the protocol throughput predictions under both symmetric and asymmetric IP traffic scenarios. Also, the proposed protocol performance is thoroughly investigated through simulation results based on Poisson and self-similar traffic model statistics, for both traffic scenarios.

This paper considers offered load estimation in probabilistic dynamic frame slotted ALOHA systems where the receiver can perceive multiple concurrent packets in a slot. This type of random access protocol finds current application in the areas of machine-to-machine communications and tag processing in RFID communications. The perception of multiple concurrent packets happens when a receiver is able to detect the number of received packets or to decode them. In the first case, we talk about multiple packet detection (MPD) whereas in the second we refer to multiple packet reception (MPR). The finite detection capability of MPD or separation capacity of MPR implies that, ideally, whenever the number of simultaneously received packets exceeds a given value M, a collision results and no information about the packets or their number can be extracted. Hence, we are confronted with an incomplete data estimation problem. We reformulate the problem so as to apply the expectation-maximization (EM) algorithm and derive the required iterative equations to obtain a maximum likelihood estimator for the offered load. A numerical example of the algorithm's execution is included and application scenarios are indicated. Simulation studies are presented to show that the proposed estimation method compares favorably with alternative techniques.

Based on software defined optical network and machine learning, the flexible resource management mechanism (ML-FRM) is proposed, which meets the resource requirements of different services in the optical data center networks. The machine learning is integrated to the SDON controller, which accomplishes the resource allocation algorithms according to the classification and clustering results. ML-FRM firstly utilizes unsupervised learning K-means algorithm to cluster traffic flows, and uses supervised learning support vector machine (SVM) algorithm to realize hierarchical classification of channel qualities. Fragmentation-Function-Fit algorithm is proposed to reduce the blocking probability, the results show that it has the lower blocking probability than First-Fit and Exact-First-Fit algorithms. ML-FRM allocates the required resources through different algorithms based on different traffic flow clustering results, and uses different modulation methods for different channel qualities. The analysis results show that ML-FRM has lower blocking probability, acceptable complexity level, and higher spectrum resource utilization efficiency than other algorithms under different offered load level.

Problem Definition and Relevance: Characterized by time-varying arrivals, multi-stage service, and multi-class patient population, emergency rooms (ERs) are complex healthcare delivery systems, where optimizing the staffing levels of physicians is a challenge. In collaboration with Mayo Clinic, we study a staffing and an associated routing problem for ER physicians and propose a new staffing rule to meet tail probability of delay (TPoD) type service targets. Methodology: We capture the time-varying patient flow in the ER with a multi-class multi-stage queuing network, describing the ER care as sequences of treatment queues, where physicians serve, and groups of diagnostic medical processes. Treatment stations in ER are busy-server queues, where waiting before service is common, but experience negligible abandonments. Motivated by these queues, we propose a new staffing algorithm, translating the offered load into staffing decisions for efficiency-driven queues with TPoD targets and perfectly patient customers. Results: We analytically show the asymptotic effectiveness of our staffing rule on stabilizing TPoD for M/M/s queues, operating in efficiency-driven mode, and numerically demonstrate its robustness and optimality in various time-varying ER settings via realistic and data-driven simulation experiments. We further show that as the service complexity of an ER increases, hybrid routing rules, using pre-determined (static) priorities and (dynamic) current system state jointly, become necessary to meet TPoD targets. Managerial Implications: Instead of treating the entire patient length of stay as a black-box service model, our multi-stage network model provides more managerial control over the internal components of ER service and is easily implementable in practice.

Fibre delay lines (FDLs) can substantially reduce the burst loss in Optical Burst Switching (OBS) networks and share-per-node FDL configurations can provide a more cost-efficient solution compared to architectures where delay lines are shared per port. Nevertheless, mathematical performance analysis of this configuration is more difficult due to traffic correlations arising from the shared resource. In this paper, an approximate two-moment traffic model is developed for quantifying end-to-end burst blocking probability in networks of OBS switches with share-per-node FDLs. The two-moment approach can improve model accuracy over more usual Poisson network analysis methods and additionally allows the characteristics of offered load to be taken into account. The accuracy of the proposed method is shown to be favourable, when compared to discrete-event simulations of an OBS network.

In this paper, we study impacts of TXOP allocation and AIFS parameter on one hop IEEE 802.11e EDCA network under varying offered load. Offered load is modified by changing the parameters of arrival process and node populations. Arrival process is varied by altering burst size distribution, mean burst size, frame size, and arrival rates. We investigate the effects of the offered load parameters on the network stability and non-saturation boundaries of IEEE 802.11e EDCA. Our results show interaction of offered load and MAC parameters. The results suggest the range of network parameters which preserve network stability and still achieve QoS differentiation.

A Fixed Point Approximation (FPA) method has recently been suggested for non-stationary analysis of loss queues and networks of loss queues with Exponential service times. Deriving exact equations relating time-dependent mean numbers of busy servers to blocking probabilities, we generalize the FPA method to loss systems with general service time distributions. These equations are combined with associated formulae for stationary analysis of loss systems in steady state through a carried load to offered load transformation. The accuracy and speed of the generalized methods are illustrated through a wide set of examples.\ud Highlights\ud \ud ► We proposed a method for performance evaluation of non-stationary loss systems. ► Our method works for single-class, multi-class and network of loss queues. ► Non-homogeneous Poisson arrivals and general service time distributions are assumed. ► The accuracy of our algorithms are shown to be close to exact. ► Our algorithms are much faster than simulation and numerical methods.

This paper presents a novel mathematical model of the IEEE 802.11 Enhanced Distributed Channel Access (EDCA) function. The main advantage of the model is that it combines accurate probability of counting down backoff slots with proper handling of frames. Furthermore, it allows for traffic differentiation with the use of traffic priority-dependent access parameters, i.e., AIFSN, and CW. Additionally, a new method of modelling AIFS differentiation between traffic classes is proposed. Finally, the model is kept reasonably simple and constructed in such a way that taking into account the four-way-handshake mechanism is uncomplicated. The proposed model is compared with simulations as well as numerical results obtained for two other models presented in the literature. The comparison gives satisfactory results regardless of the offered load, number of nodes, or network configuration.

In the new era of Ambient Intelligence, wireless sensor networks (WSNs) are seen to bridge the gap between physical world and the Internet, making a large amount of information accessible anywhere, anytime. Over the last few years, WSNs are being developed towards a large number of multimedia streaming applications, e.g. video surveillance, traffic control systems, health monitoring, and industrial process control. WSNs consist of small sensor devices (nodes) that are capable of working unattended, without centralized control, under dynamically changing conditions. However, these devices face important limitations in terms of energy, memory and computational power. The uncontrolled use of limited resources in conjunction with the unpredictable nature of WSNs in terms of traffic load injection, wireless link capacity fluctuations and topology modifications may lead to congestion. Congestion can cause increased packet loss and delay. This paper proposes a bio-inspired congestion control approach for WSNs streaming applications that necessitate controlled performance with graceful degradation. In the proposed approach, congestion in WSNs is prevented (or at least minimized) by regulating the rate of each traffic flow based on the Lotka-Volterra competition model. Performance evaluations reveal that the proposed approach achieves adaptability to changing traffic loads, scalability and fairness among flows, while providing graceful performance degradation as the offered load increases. © 2010 Elsevier B.V. All rights reserved. 33 17 2039 2047 Cited By :20

In this paper we model and study the performance of peer-to-peer (P2P) file sharing systems in terms of their 'service capacity'. We identify two regimes of interest: the transient and stationary regimes. We show that in both regimes, the performance of P2P systems exhibits a favorable scaling with the offered load. P2P systems achieve this by efficiently leveraging the service capacity of other peers, who possibly are concurrently downloading the same file. Therefore to improve the performance, it is important to design mechanisms to give peers incentives for sharing/cooperation. One approach is to introduce mechanisms for resource allocation that are 'fair', such that a peer's performance improves with his contributions. We find that some intuitive 'fairness' notions may unexpectedly lead to 'unfair' allocations, which do not provide the right incentives for peers. Thus, implementation of P2P systems may want to compromise the degree of 'fairness' in favor of maintaining system robustness and reducing overheads.

Strict priority scheme is one of several ways for the implementation of a service differentiation in a router because of simplicity in implementation and its capability to guarantee the delay requirement of the highest-class of traffic. It is also known that strict priority to highest-priority class traffic starves the lower-classes traffic at the cost of prioritizing the highest-class traffic. Starvation of lower-class traffic may be problematic in case the offered load of highest-class traffic is high. However, under some proper mixing of the traffic and over a high-speed link, the concerns for the starvation problem may not be so serious. In addition, we argue that delay of the highest-class traffic can be influenced by the behaviors of the lower-class traffic even under the strict priority service scheme. The purpose of this work is to present an analytic method which can represent the delay performance of the strict priority scheme in an explicit manner from those two viewpoints, and shows that it may be applied to a service differentiation scheme for the future IP network with low or moderate level of the highest-class traffic. Via numerical experiments we validate our argument.

Recent researches present that packet-mode scheduling may provide overall performance advantages over cell-mode scheduling in input queueing switches. But one significant drawback of existing packet-mode schedulers is that they tend to result in longer waiting time of short packets, whose delay is critical to the performance of IP networks. To mitigate the problem, this paper proposes a simple but effective packet-mode scheduling algorithm called Short Packets First (SPF), studies the corresponding buffering architecture, and evaluates the performance on throughput and packet delay. SPF achieves 100% throughput under an accurate Internet traffic model. Theoretical analysis proves that SPF can reduce the average packet waiting time of short packets as well as overall packets, under the condition of uniform Poisson arrival process and low to medium offered load. Furthermore, extended simulation studies show that SPF is superior to general packet-mode and cell-mode scheduling, especially under heavy offered load.

In this paper, we investigate the performance of routing and rate allocation (RRA) algorithms in rate-based multi-class networks. On the arrival of a connection request, an RRA algorithm selects a route for the connection and allocates an appropriate rate on the route; failing this, it blocks the connection request. We measure the performance of an RRA algorithm in terms of its minimum weighted carried traffic. This performance criterion encompasses two widely used performance criteria, namely, weighted carried traffic and minimum carried traffic. We derive an upper bound on the minimum weighted carried traffic of any RRA algorithm. The bound can be computed by solving a linear program. Moreover, we show that the bound is achieved asymptotically, when the offered load and the link capacities are large, by a Partitioning RRA algorithm. Therefore the bound can be used as a performance benchmark for any RRA algorithm. We observe that the proposed Partitioning RRA algorithm, though asymptotically optimal, performs poorly at very low loads. We investigate the cause of this undesirable behaviour and obtain two improved asymptotically optimal RRA algorithms.

Multi-protocol label switching extends the IP destination-based routing protocols to provide new and scalable routing capabilities in connectionless networks using relatively simple packet forwarding mechanisms. MPLS networks carry traffic on virtual connections called label switched paths. This paper considers path selection and bandwidth allocation in MPLS networks in order to optimize the network quality of service. The optimization is based upon the minimization of a non-linear objective function which under light load simplifies to OSPF routing with link metrics equal to the link propagation delays. The behavior under heavy load depends on the choice of certain parameters. It can essentially be made to minimize maximal expected utilization, or to maximize minimal expected weighted slacks (both over all links). Under certain circumstances it can be made to minimize the probability that a link has an instantaneous offered load larger than its transmission capacity. We present a model of an MPLS network and an algorithm which optimally distributes the traffic among a set of active paths and reserves a set of back-up paths for carrying the traffic of failed or congested paths. The algorithm is an improvement of the well-known flow deviation non-linear programming method. The algorithm is applied to compute optimal LSPs for a 100-node network carrying a single traffic class. A link carrying some 1400 routes fails. The back-up paths are activated and we compare the performance of the path sets before and after the back-up paths are deployed.

This article investigates the unfairness and throughput deterioration phenomena in distributed queue dual bus (DQDB) with slot reuse (DQDB-SR), and then proposes a bandwidth control method to prevent the phenomena. DQDB-SR is generally implemented by introducing erasure nodes which divide the network into multiple segments. When the offered load exceeds the maximum throughput level of DQDB-SR, requests from downstream segments begin to flow towards their upstream segments. This request overflow incurs a segment unfairness which forces stations on the upstream segments to be starved in access to the network. The efficiency of slot reuse will then decline, which leads to a throughput deterioration. We apply the bandwidth balancing mechanism and the access protection scheme (APS) used in DQDB to overcome these problems, and compare their performance under various workload types. Simulation results show that the APS is quite suitable for the bandwidth control to improve the performance of DQDB-SR at overload conditions.

In an Ethernet with high offered load, lengthy and highly variable packet delays can occur. These delays are due in part to the ''capture effect'' whereby a single station can dominate packet transmission on an Ethernet that uses the standard Binary Exponential Backoff (BEB) arbitration algorithm despite other stations also having packets queued for transmission. One proposed solution to this problem is the Binary Logarithmic Arbitration Method (BLAM) under review by the IEEE 802.3 standards committee. This paper presents a performance evaluation of BLAM and a new, simpler arbitration method called BEB2. It is shown that both BLAM and BEB2 reduce the mean and variance of packet delay and can thus improve the transport of real-time voice traffic on Ethernet.

A dominant factor for the widespread penetration and success of B-ISDN is the access cost for the low traffic customers. For residential and small business customers, the tree Passive Optical Networks (PONs) are a promising solution for the concentration of ATM traffic. Such a solution however requires a method to arbitrate the access to the common Line Termination using the shared slotted system created on the PON tree. The Medium Access Control (MAC) protocol presented in this work employs delay priorities to achieve more efficient multiplexing with better jitter performance in the multi-service B-ISDN environment. It respects the demanding traffic control and enforcement methods of ATM which do not tolerate certain traffic profile distortions. It avoids cell clustering which deteriorates policing efficacy and link utilisation. It shields time-constrained services from the adverse effect on their traffic profile of the contention from bursty services so that they see a very low offered load. The delay insensitive ABR cells are buffered and filled-in when traffic conditions allow on a best-effort basis. This results in a more intelligent multiplexing and smoother traffic stream at the exit. Thus a higher loading is possible for the same performance leading to 'amortisation' of the additional complexity due to prioritisation.

In this paper, assuming two classes of input traffic, we analyze a combination of push-out and general state-dependent scheduling schemes applied to input queues of nonblocking ATM switches with the speed-up factor ≥ 1. The analysis is done in two steps in the discrete-time domain, using an embedded semi-Markov process (SMP) concept (Takagi, 1993). In numerical examples for some specific scheduling schemes, we compare the performances and show that reasonable performance differences both in loss probabilities and in waiting times can be achieved between two classes over a wide range of the offered load. We also show that all the priority schemes considered in this paper allow higher utilization of the switch than the non priority one, and among them the queue length threshold (QLT) scheduling scheme gets the highest utilization. Lastly, we note that although the analysis assumes an infinitely large size switch, the result gives a quite accurate approximation for finite size switches with around one hundred input/output ports.

The MetaRing is a medium access control (MAC) protocol for gigabit LANs and MANs with cells removed by the destination stations (slot reuse). Slot reuse increases the aggregate throughput beyond the capacity of single links but may cause starvation. In order to prevent this the MetaRing MAC protocol includes a fairness mechanism. Two types of fairness algorithms have been proposed: `global' and `local'. The MetaRing analysed in this paper implements the local fairness algorithm specified in [2]. In order to reduce the complexity we have identified a simplified model which can be analytically solved and yet still provides useful information on network performances. This model represents a worst-case scenario in which network congestion is stressed, i.e. no station, apart from a specific station (tagged station), ever has an empty queue. The model proposed can be represented by a discrete time discrete state Markov chain of M/G/1-type and hence the matrix analytical methodology has been used to solve it. Our analysis focuses on the average access delay experienced by the tagged station as a function of the offered load. The results show that average access delay remains bounded for values of offered load less than or equal to 90%. Furthermore, the average access delay depends on the number of interfering stations and on the protocol parameter Q"r.

The p"i-persistent medium access protocol is an attractive solution for high-speed time-slotted unidirectional bus networks. This protocol, with its very simple flow control mechanism allowing station i to access empty slots with station-dependent probability p"i, has been seen as an alternative for solving fairness problems associated with the original IEEE 802.6 standard for the distributed queue dual bus (DQDB) for high-speed networks. In this paper we analyze the main aspects of the fairness provision mechanism employed in the p"i-persistent protocol. Assuming heavy traffic load, the stations' access probabilities are determined so as to allocate certain portions of the channel capacity for guaranteeing prescribed service requirements. Having determined the probability generating function of queue length as well as the two first moments of the worst-case packet delay (delay of the last packet arriving at a station within a slot time) for a single station, we investigate packet delays and fairness in a bus network with N stations. Our results show that, in a network with homogeneous Poissonian arrival streams, if the bus is able to carry all the offered load, protecting fairness of services by lowering the access probability at a station leads only to deterioration of that station's performance without affecting performance of other stations. This is a surprising result which however does not apply for bursty traffic, as shown by the results of our simulation studies.

In a recent study, we found that average demand for bandwidth in the recent past, and also in the future according to reasonable projections such as the Cisco VNI, is far less than most analysts have assumed. Even in 2020, average peak hour demand per broadband/NGA user can be expected to be not more than 2 Mbps. This finding would appear to represent very good news for the Digital Agenda for Europe (DAE), and in particular would seem to imply that there is more scope than many have assumed for network infrastructures such as DOCSIS-enabled cable and LTE that depend significantly on shared bandwidth to meet realistic DAE needs for 30 Mbps and even 100 Mbps ultra-fast broadband service well into the future.The finding also raises complex questions, however, about the nature of consumer demand for broadband bandwidth. It is clear that lack of bandwidth can suppress consumption of network bandwidth. One might well imagine that consumer bandwidth consumption is largely independent of the speed of network access, as long as the capacity exceeds the demand ("offered load", as a queuing theorist would call it). Overall, this view would hold that too little bandwidth can reduce consumption, but that "too much" bandwidth availability cannot promote bandwidth consumption.Some have argued that bandwidth plays a role well beyond this – that the presence of high bandwidth access promotes creation of new services, and thus consumption of whatever bandwidth is available. This alternative view is analogous to the common observation that no matter how many closets a house has, they tend to quickly be filled up. The difference is important for policymakers, and cannot be resolved by abstract reasoning alone. This is an empirical question.We have conducted a preliminary assessment that, while not definitive, strongly suggests that whatever "build it and they will come" effect might exist cannot be very strong. This in turn poses troubling policy questions as regards public policy to promote the deployment of ultrafast broadband.

Several important modifications to the distributed queue dual bus (DQDB) (IEEE 802.6) protocol for metropolitan area networks (MANs) are proposed. These modifications are aimed at improving not only the throughout and delay performance of the network but also its fairness performance. The first proposal exploits the continuation-of-message (COM) information in a DQDB segment in order to minimize the mean overhead for multi-segment messages. Then, a scheme for combining the COM information along with the spatial reuse of slots is proposed, thereby achieving even higher throughput and lower delay. Simulation models are used to study the performance gains achievable by using the COM and the slot reuse approaches. Although slot reuse has been proposed before to improve the DQDB protocol's performance, DQDB's unfairness problem becomes more pronounced when slots are reused. This paper also proposes several schemes to achieve different fairness criteria in a DQDB network employing slot reuse, namely: (a) an individual station's throughput proportional to the number of stations downstream from itself, (b) equal throughput at all stations, and (c) an individual station's throughput proportional to its own offered load. Analytical results corresponding to these schemes are developed, and they are verified via simulation. The proposed schemes can be implemented with just one additional counter and one register (on top of the regular DQDB hardware) at each station.

In many processing systems, the goodput (the number of succesfully completed tasks per time unit) may decrease as the offered load increases. Such systems need some form of overload control. We distinguish between open-loop control that limits the number of tasks in the system with a fixed maximum and closed-loop control that rejects new tasks if the observed processor loading is too high. In most control systems closed-loop control performs better than open-loop control. Here we present a remarkable illustration of the converse: we show that higher goodput can be obtained with an open-loop control-law. The results are obtained with a simplified analytical model of a multi-tasking processing system, such as the central processor of a telephone exchange. Compared to open-loop control, closed-loop control is more robust with respect to a change in task idle-time. Under conditions of overload, the mentioned differences between the two types of control law become more pronounced. The computed results are confirmed by the outcome of simulation runs. In the simulations we used a more realistic definition of goodput than was used in the analytical model.

A family of local area network protocols is presented. The LAN we consider consists of a ring and a bidirectional bus to which stations are connected via adapters and passive taps respectively. The data frames in these protocols can avoid several stations resulting in less delays. The main feature of these protocols is that reservation procedures for the bypassed stations take place in parallel with the data frame transmission. Another interesting feature is that all the standard token ring functions do operate under the proposed topology. The last member of this family is a protocol with the ability to change its behaviour according to the offered load at each station. The performance of the protocols is evaluated by simulation models. Also performance comparisons of the new protocols are presented.

In this paper, the service times for two types of the user request on the Mini-MAP network are analyzed for the structures of the MMSE(Manufacturing Messaging Service Element) implementation. One structure has the MMSE lying in the NIU(Network Interface Unit) and the other in the host system. A processing time generated by the MMSE and the data link layer is considered for the analysis. The results obtained show that the service time depends heavily on the structures of the MMSE implementation and the processing time. Specially, when the sum of the processing times to encode and to decode does not exceed about 10msec, the structure that the MMSE lies in the NIU has an advantage in that the offered load to the host decreases while the service time is within 25msec. Also, if the explicit association has to be used for the communication, the fast responsiveness, which is the main requirement of Mini-MAP, is lost for both structures even though the processing time in the NIU is small.

A medium access protocol for a shared, multiaccess bus network is proposed. The principles of this access method, called Segnet, are described. A simulation model for studying the performance of this protocol has been developed, and the results of the simulation show that Segnet performs significantly better than CSMA/CD when the offered load to the network is high.

We consider a system which supports a load in such a way that the profit earned by the system for one time unit is a continuous and increasing function of the amount of the load. The time-to-failure distribution of the system is assumed to be gamma. Both parameters of the distribution are continuous functions of the load such that the mean time-to-failure is a decreasing function of the load. The system earns the profit only when it is on. When it is off, it is repaired, and the losses are assumed to be proportional to the repair time. We define the average asymptotic gain earned by the system in a single time unit; this quantity can be thought of as a system efficiency measure. Assuming that the load is a controllable variable, the goal is to find a value of the load that maximizes the so-defined efficiency. To model the relationship between the load and the parameters of the time-to-failure distribution (gamma) as well as the relation between the load and the profit, power and exponential functions have been taken. The analytical conditions for the existence of maximum positive gain and the analysis of the behaviour of the gain as the load varies are presented.

Local area networks are becoming increasingly important in the design of computer systems. It is therefore important that all aspects of their performance are understood. This paper presents a performance comparison, using simulation techniques, of two leading examples of the genre: 1. 1. The Cambridge Digital Communication Ring: a point to point communications ring using the empty slot principle for statistical multiplexing. 2. 2. Ethernet: a branching broadcasting system using the CSMA-CD protocol. The paper also briefly discusses other factors involved in a complete cost/benefit analysis.

We present an efficient method for computing the steady state probabilities of a finite capacity queue with exponential service. Poisson arrivals, non-preemptive or preemptive-resume priority scheduling, and common or reserved waiting areas for high and low priority packets. Numerical results are presented for the complete sharing (CS), complete partition (CP), sharing with minimum allocation (SMA) and sharing with maximum queue length (SMXQ) buffer allocation schemes with two priority classes. Under complete sharing, the admitted arrival rates are independent of the priority ordering and the queue lengths and response (sojourn) times of the admitted packets need not be monotonic or even increasing in the offered loads. Under CP, SMA and SMXQ, low priority packets may be subject to starvation. Under all schemes considered, the response times of either priority classes may decrease as the total offered load is increased.

The modeling and optimal flow control of a Jacksonian network in equilibrium is investigated. The model employed consists of a controller node cascaded with the Jacksonian network. Input packets arrive at the controller node with a Poissonian rate δ. For a blocking type strategy for accessing the network it is shown that the control which maximizes the average throughput of the network subject to a bounded average time delay constraint is a window flow control mechansim. The window size depends on the offered load δ, the maximum service rate of the controlling queueing system, c , and the Norton equivalent service rate of the network μ. The dependence of the average throughput and the average time delay on the control is also analyzed.

Load indices that accurately reflect the current loads at computer system resources are crucial to the success of any dynamic load balancing scheme. However, few load indices have been experimentally validated as being suitable for load balancing. We conduct such a validation study for the resource queue lengths. We find that the CPU and disk queue lengths have very high correlation to the response times of CPU and I/O bound jobs, respectively. However, for the type of system that we studied, the system load changes very rapidly, making the response time high unpredictable. Simulation results suggest that load balancing will drastically reduce load fluctuation. The CPU is by far the most heavily used resource in the systems we studied. While this is also true in many other environments, measurement studies are called for before reaching such a conclusion in other systems