Average Queue Size Research Articles

Periodic GATE Optimization (PGO): A New Service Scheme for Long-Reach Passive Optical Networks

In this paper, we present a new service scheme, Periodic GATE Optimization (PGO) for long-reach passive optical networks. PGO is based on a previously proposed service scheme where the optical line terminal (OLT) generates multiple threads to poll the bandwidth requests of the optical network units (ONUs). In PGO, OLT periodically builds an ILP formulation by using the collected requests of the ONUs, and solves the model for the overloaded ONUs. Based on the outputs, it determines how to credit the ONUs until the next ILP formulation whenever they are overloaded. We evaluate the performance of PGO by simulation, and show that it introduces further decrease in average packet delay until the heavy loads. Average queue length at the ONUs is also shortened by PGO. The results imply that PGO achieves these enhancements without increasing the packet drop probability. In the paper, we also present the quality-of-service (QoS) aware version of PGO, namely PGO-QoS. PGO-QoS consists of two modules such as intra-ONU scheduling where the dequeuing proportion of the QoS classes is determined and reported to OLT, and dynamic bandwidth allocation module where mostly PGO runs. We evaluate the performance of PGO-QoS for each QoS class and for the overall network in terms of average packet delay, average queue size, and packet drop probability under different scenarios. Based on the simulation results, PGO-QoS is shown to introduce further decrease in delay, queue length, and packet loss probability for the high priority class requests that are granted by multithread polling in Long-Reach PON.

Self‐Tuning Random Early Detection Algorithm to Improve Performance of Network Transmission

We use a discrete‐time dynamical feedback system model of TCP/RED to study the performance of Random Early Detection (RED) for different values of control parameters. Our analysis shows that the queue length is able to keep stable at a given target if the maximum probability pmax and exponential averaging weight w satisfy some conditions. From the mathematical analysis, a new self‐tuning RED is proposed to improve the performance of TCP‐RED network. The appropriate pmax is dynamically obtained according to history information of both pmax and the average queue size in a period of time. And w is properly chosen according to a linear stability condition of the average queue length. From simulations with ns‐2, it is found that the self‐tuning RED is more robust to stabilize queue length in terms of less deviation from the target and smaller fluctuation amplitude, compared to adaptive RED, Random Early Marking (REM), and Proportional‐Integral (PI) controller.

SYMBOLIC DYNAMICAL MODEL OF AVERAGE QUEUE SIZE OF RANDOM EARLY DETECTION ALGORITHM

In this paper, a symbolic dynamical model of the average queue size of the random early detection (RED) algorithm is proposed. The conditions for both the system parameters and the initial conditions that the average queue size of the RED algorithm would converge to a fixed point are derived. These results are useful for network engineers to design both the system parameters and the initial conditions so that internet networks can achieve a good performance.

Random Early Detection for Congestion Avoidance in Wired Networks: A Discretized Pursuit Learning-Automata-Like Solution

In this paper, we present a learning-automata-like The reason why the mechanism is not a pure LA, but rather why it yet mimics one, will be clarified in the body of this paper. (LAL) mechanism for congestion avoidance in wired networks. Our algorithm, named as LAL Random Early Detection (LALRED), is founded on the principles of the operations of existing RED congestion-avoidance mechanisms, augmented with a LAL philosophy. The primary objective of LALRED is to optimize the value of the average size of the queue used for congestion avoidance and to consequently reduce the total loss of packets at the queue. We attempt to achieve this by stationing a LAL algorithm at the gateways and by discretizing the probabilities of the corresponding actions of the congestion-avoidance algorithm. At every time instant, the LAL scheme, in turn, chooses the action that possesses the maximal ratio between the number of times the chosen action is rewarded and the number of times that it has been chosen. In LALRED, we simultaneously increase the likelihood of the scheme converging to the action, which minimizes the number of packet drops at the gateway. Our approach helps to improve the performance of congestion avoidance by adaptively minimizing the queue-loss rate and the average queue size. Simulation results obtained using NS2 establish the improved performance of LALRED over the traditional RED methods which were chosen as the benchmarks for performance comparison purposes.

The mechanism of adapting RED parameters to TCP traffic

Random Early Detection (RED) can stabilize the instantaneous queue size at a router by controlling the average queue size within a given target. In this case, RED can achieve high throughput in the routers. However, the average queue size is quite sensitive to network scenarios and it is difficult to adapt RED parameters to changing network traffic. In this paper we use a previously developed dynamic model of TCP behavior together with a linear feedback model of TCP/RED to analyze and design a mechanism for RED parameter tuning in response to changing network conditions like traffic load, link capacity and round-trip time. Even though the values of four key RED parameters are determined by varying network conditions, they can be tuned independently without consideration of the interactions among these RED parameters. Simulation results show that this mechanism can keep the instantaneous queue size stable and maintain high link utilization in a wide variety of network conditions.

On the Scalable Fairness and Efficient Active Queue Management of RED

Internet routers generally see packets from a fast flow more often than a slow flow. This suggests that network fairness may be improved without per-flow information. In this paper, we propose a scheme using Most Recently Used List (MRUL)-a list storing statistics of limited active flows that sorted in most recently seen first mode-to improve the fairness of RED. Based on the list, our proposed scheme jointly considers the identification and punish of the fast and unresponsive fast flows, and the protection of slow flows. Its performance improvements are demonstrated with extensive simulations. Different from the previous proposals, the complexity of our proposed scheme is proportional to the size of the MRUL list but not coupled with the queue buffer size or the number of active flows, so it is scalable and suitable for various routers. In addition, another issue we address in this paper is queue management in RED. Specifically, we replace the linear packet dropping function in RED by a judicially designed nonlinear quadratic function, while original RED remains unchanged. We call this new scheme Nonlinear RED, or NLRED. The underlying idea is that, with the proposed nonlinear packet dropping function, packet dropping becomes gentler than RED at light traffic load but more aggressive at heavy load. As a result, at light traffic load, NLRED encourages the router to operate in a range of average queue sizes rather than a fixed one. When the load is heavy and the average queue size approaches the pre-determined maximum threshold (i.e. the queue size may soon get out of control), NLRED allows more aggressive packet dropping to back off from it. Simulations demonstrate that NLRED achieves a higher and more stable throughput than RED and REM. Since NLRED is fully compatible with RED, we can easily upgrade/replace the existing RED implementations by NLRED.

Bin Packing with Queues

We study the best achievable performance (in terms of the average queue size and delay) in a stochastic and dynamic version of the bin-packing problem. Items arrive to a queue according to a Poisson process with rate 2ρ, where ρ ∈ (0, 1). The item sizes are independent and identically distributed (i.i.d.) with a uniform distribution in [0, 1]. At each time unit, a single unit-size bin is available and can receive any of the queued items, as long as their total size does not exceed 1. Coffman and Stolyar (1999) and Gamarnik (2004) have established that there exist packing policies under which the average queue size is finite for every ρ ∈ (0, 1). In this paper we study the precise scaling of the average queue size, as a function of ρ, with emphasis on the critical regime where ρ approaches 1. Standard results on the probabilistic (but static) bin-packing problem can be readily applied to produce policies under which the queue size scales as O(h2), where h = 1 / (1 - ρ), which raises the question of whether this is the best possible. We establish that the average queue size scales as Ω(hlogh), under any policy. Furthermore, we provide an easily implementable policy, which packs at most two items per bin. Under that policy, the average queue size scales as O(hlog3/2h), which is nearly optimal. On the other hand, if we impose the additional requirement that any two items packed together must have near-complementary sizes (in a sense to be made precise), we show that the average queue size must scale as Θ(h2).

Bin Packing with Queues

We study the best achievable performance (in terms of the average queue size and delay) in a stochastic and dynamic version of the bin-packing problem. Items arrive to a queue according to a Poisson process with rate 2ρ, where ρ ∈ (0, 1). The item sizes are independent and identically distributed (i.i.d.) with a uniform distribution in [0, 1]. At each time unit, a single unit-size bin is available and can receive any of the queued items, as long as their total size does not exceed 1. Coffman and Stolyar (1999) and Gamarnik (2004) have established that there exist packing policies under which the average queue size is finite for every ρ ∈ (0, 1). In this paper we study the precise scaling of the average queue size, as a function of ρ, with emphasis on the critical regime where ρ approaches 1. Standard results on the probabilistic (but static) bin-packing problem can be readily applied to produce policies under which the queue size scales as O(h 2), where h = 1 / (1 - ρ), which raises the question of whether this is the best possible. We establish that the average queue size scales as Ω(hlogh), under any policy. Furthermore, we provide an easily implementable policy, which packs at most two items per bin. Under that policy, the average queue size scales as O(hlog3/2 h), which is nearly optimal. On the other hand, if we impose the additional requirement that any two items packed together must have near-complementary sizes (in a sense to be made precise), we show that the average queue size must scale as Θ(h 2).

Static and dynamic server allocation in systems with on/off sources

A system consisting of a number of servers, where demands of different types arrive in bursts (modelled by interrupted Poisson processes), is examined in the steady state. The problem is to decide how many servers to allocate to each job type, so as to minimize a cost function expressed in terms of average queue sizes. First, an exact analysis is provided for an isolated IPP/M/n queue. The results are used to compute the optimal static server allocation policy. The latter is then compared to four heuristic policies which employ dynamic switching of servers from one queue to another (such switches take time and hence incur costs).

ANALYSIS AND CONTROL OF BIFURCATION AND CHAOS IN AVERAGE QUEUE LENGTH IN TCP/RED MODEL

This paper studies the bifurcation and chaos phenomena in average queue length in a developed Transmission Control Protocol (TCP) model with Random Early Detection (RED) mechanism. Bifurcation and chaos phenomena are nonlinear behavior in network systems that lead to degradation of the network performance. The TCP/RED model used is a model validated previously. In our study, only the average queue size [Formula: see text] is considered, and the results are based on analytical model rather than actual measurements. The instabilities in the model are studied numerically using the conventional nonlinear bifurcation analysis. Extending from this bifurcation analysis, a modified RED algorithm is derived to prevent the observed bifurcation and chaos regardless of the selected parameters. Our modification is for the simple scenario of a single RED router carrying only TCP traffic. The algorithm neither compromises the throughput nor the average queuing delay of the system.

Effective RED: An algorithm to improve RED's performance by reducing packet loss rate

Random early detection (RED) is an effective congestion control mechanism acting on the intermediate gateways. We describe a new active queue management scheme, Effective RED (ERED) that aims to reduce packet loss rates in a simple and scalable manner. We made a few change to packet drop function of existing RED scheme. The rest of the original RED remains unchanged. We achieve this by making several refinements and by controlling packet dropping function both with average queue size and instantaneous queue size. Simulations demonstrate that ERED achieves a highest throughput and lowest packet drops than RED, Blue, REM, FRED, LDC and SRED. Since ERED is fully compatible with RED, we can easily upgrade/replace the existing RED implementations by ERED.

Simulation comparison of active queue management algorithms in TCP/IP networks

Active Queue Management techniques are recommended to overcome the performance limitations of TCP congestion control mechanisms over drop-tail networks. The main objective of this paper is to present the comparative analysis of the performance of 10 different queue management policies using the OMNeT++ simulator. The AQMs have many parameters which influence their behavior, so some results concerning the parameters selection are given. The algorithms are tested in terms of average queue size, packet delay and packet loss rate in presence of TCP traffic.

Comparative study of stability in different TCP/RED models

This paper studies the stability issue of the average queue length of a Transmission Control Protocol (TCP) model when interacting with Random Early Detection (RED). The model used for the study has shown period doubling bifurcation (PDB) and border collision bifurcation (BCB) in the average queue size at certain values of parameters when original RED is deployed. In this paper, we adopt a gentle version of RED and a newly derived RED algorithm into the model to study the improvement in stability of average queue size of the system.

Hybrid control of bifurcation and chaos in stroboscopic model of Internet congestion control system

Interaction between transmission control protocol (TCP) and random early detection (RED) gateway in the Internet congestion control system has been modelled as a discrete-time dynamic system which exhibits complex bifurcating and chaotic behaviours. In this paper, a hybrid control strategy using both state feedback and parameter perturbation is employed to control the bifurcation and stabilize the chaotic orbits embedded in this discrete-time dynamic system of TCP/RED. Theoretical analysis and numerical simulations show that the bifurcation is delayed and the chaotic orbits are stabilized to a fixed point, which reliably achieves a stable average queue size in an extended range of parameters and even completely eliminates the chaotic behaviour in a particular range of parameters. Therefore it is possible to decrease the sensitivity of RED to parameters. By using the hybrid strategy, we may improve the stability and performance of TCP/RED congestion control system significantly.

Reduction of queue oscillation in the next generation Internet routers

The Internet routers employing the random early detection (RED) algorithm for congestion control suffer from the problem of chaotic queue oscillation. It is well known that the slowly varying nature of the average queue size computed using an exponentially weighted moving average (EWMA) used in the RED scheme causes this chaotic behavior. This paper presents a new mathematical function to model the weighting parameter used in the EWMA. The proposed weighting function incorporates the knowledge of the dynamic changes in the congestion characteristics, traffic characteristics and queue normalization. Using this pragmatic information eliminates the slowly varying nature of the average queue size. It is evident from our simulations that the proposed approach not only reduces the chaotic queue oscillation significantly but also provides predictable low delay and low delay jitter with high throughput gain and reduced packet loss rate even under heavy load of traffic conditions.

Random Early Detection method for ABR service

A new implementation for the Random Early Detection method algorithm for ABR (REDM–ABR) service is proposed in this paper. It keeps running an exponential average of the queue length ( Q). When a cell arrives, the average queue size ( Q avg) is compared with two threshold levels, lower queue threshold ( Q L) and higher queue threshold ( Q H). If it is smaller than Q L, the cell is passed, but if it is larger than Q H, the cell marking probability is set to one. If it is in between the two thresholds, the cell marking probability is calculated depending on the value of Q avg. The values of Resource Management (RM) cell fields: Congestion Indication (CI), No Additive Increase (NI), and Explicit Rate (ER) are filled by the Relative Rate Marking (RRM) switch and sent back to the sources. The sources will change their rate depending on CI, NI, and ER values. To investigate the effect of decreasing the impending congestion area, a dynamic Q threshold ( Q D) is used. The Q D is shifted from Q L toward Q H for decreasing the congestion area and to investigate the effect of this shift on the performance of the switch through simulation study.

Performance Analysis of IPACT Media Access Control Protocols for Gigabit Ethernet-PONs

In this paper, we examine the Interleaved Polling with Adaptive Cycle Time (IPACT) that was proposed to control upstream traffic for Gigabit Ethemet-PONs, a promising technology for the Fiber To The Home (FTTH). We analyzed the performance for the gated service and the limited service mathematically. To do this, the IPACT protocol was modeled as a polling system and analyzed by using mean-value analysis technique. The traffic arrival rate A was divided into three regions, and each region was analyzed separately and merged appropriately by using an interpolation method. The average packet delay, average queue size, and average cycle time of both the gated service and the limited service were obtained through the analysis. In order to evaluate the accuracy of the mathematical analysis, discrete event simulation was performed for the IPACT protocol. Simulation results show the accuracy of the mathematical analysis. The analysis results can be widely used in the design of the FTTH system based on EPON, as the performance results in the present study can be obtained in a rather short time. We can design an appropriate system depending on various traffic conditions by adjusting system parameters, such as the number of users N, the maximum transfer window W MAX , and so on.

A proactive mechanism for quality of service control in high speed networks

The internet offers best effort service only. The problem of providing Quality of Service (QoS) control to multimedia flows in a congested router is challenging in this scenario. A novel Active Queue Management (AQM) mechanism is proposed for use at the router. The design and modelling of this queue-based AQM is presented. The buffer at the outgoing link is a simple First-In-First-Out, shared by packets belonging to n flows. A packet-dropping scheme with a single threshold is applied. As against Random Early Detection (RED), which uses the average queue size to control dropping, this work uses instantaneous queue size for making the decisions. This improves the buffer utilisation at the IP routers. The mechanism helps multimedia applications by giving preferential treatment to their flows. The mechanism also considers inter-protocol fairness by reducing loss rates of Transmission Control Protocol (TCP) flows. By doing this, the scheme implements the fair queuing policy. The scheme is stateless and easy to implement. Simulation results indicate that the new service model gives a better than best-effort treatment to the flows that require good QoS.

Opportunistic medium access for wireless networking adapted to decentralized CSI

Relative to a centralized operation, opportunistic medium access capitalizing on decentralized multiuser diversity in a channel-aware homogeneous slotted Aloha system with analog-amplitude channels has been shown to incur only partial loss in throughput due to contention. In this context, we provide sufficient conditions for stability as well as upper bounds on average queue sizes, and address three equally important questions. The first one is whether there exist decentralized scheduling algorithms for homogeneous users with higher throughputs than available ones. We prove that binary scheduling maximizes the sum-throughput. The second issue pertains to heterogeneous systems where users may have different channel statistics. Here we establish that binary scheduling not only maximizes the sum of the logs of the average throughputs, but also asymptotically guarantees fairness among users. The last issue we address is extending the results to finite state Markov chain (FSMC) channels. We provide a convex formulation of the corresponding throughput optimization problem, and derive a simple binary-like access strategy

Nonlinear RED: A simple yet efficient active queue management scheme

Among various active queue management schemes (AQM), random early detection (RED) is probably the most extensively studied. Unlike the existing RED enhancement schemes, we replace the linear packet dropping function in RED by a judicially designed nonlinear quadratic function. The rest of the original RED remains unchanged. We call this new scheme Nonlinear RED, or NLRED. The underlying idea is that, with the proposed nonlinear packet dropping function, packet dropping becomes gentler than RED at light traffic load but more aggressive at heavy load. As a result, at light traffic load, NLRED encourages the router to operate in a range of average queue sizes rather than a fixed one. When the load is heavy and the average queue size approaches the pre-determined maximum threshold (i.e. the queue size may soon get out of control), NLRED allows more aggressive packet dropping to back off from it. Simulations demonstrate that NLRED achieves a higher and more stable throughput than RED and REM, another efficient variant of RED. Since NLRED is fully compatible with RED, we can easily upgrade/replace the existing RED implementations by NLRED.