James Yan

Analysis of SIP retransmission probability using a Markov-Modulated Poisson Process model

As a main signaling protocol for multimedia sessions in the Internet, SIP (Session Initiation Protocol) introduces a retransmission mechanism to maintain the reliability for its real-time transmission. However, retransmission will make the server overload worse. Recent collapse of SIP servers due to emergency-induced call volume indicates that the built-in SIP overload control mechanism cannot prevent the server from overload collapse under heavy load. In this paper, we apply a MMPP (Markov-Modulated Poisson Process) model to analyze the queuing mechanism of SIP server under two typical service states. The MMPP model allows us to investigate the probability of SIP retransmissions. By performing numerous experiments statistically to verify SIP retransmission probability calculated by MMPP model, we find that high retransmission probability caused by short demand surge or reduced server processing capacity during maintenance period may overload and crash a server. We run simulations using time-series directly to observe and analyze the system performance of an overloaded SIP server. This is much faster than event-driven simulation. Numerical results demonstrate that low resource utilization corresponds to low retransmission probability. However, a utilization as low as 20% cannot always guarantee a SIP system stability upon a temporal server slowdown or a short period of demand burst.

IMS network deployment cost optimization based on flow-based traffic model

The IP Multimedia Subsystem (IMS) is envisioned as the next-generation IP-based multimedia system that integrates data, speech, and video network services over both wireless and wireline networks. In this paper, we focus on optimizing the cost of SIP server deployment in an IMS network. To reflect the traffic loads on the servers, a flow-based model is used to characterize the SIP traffic. Formulated as a linear programming problem, the cost optimization involves mapping a logical IMS core network topology into a physical network topology. Three potential mapping strategies are proposed. Each strategys specific constraints are incorporated into the mathematical formulation of the problem. A numerical example of each strategy is presented, and the discussion on the formulations is provided.

Adaptive sampling for network performance measurement under voice traffic

As the Internet grows in scale and complexity, the benefits of network performance measurements and monitoring are significantly increasing. Sampling-based measurement methods provide adequate techniques for reducing the quantity of control data and attract growing interests. The research reported in this paper, addresses the issue of how to carry out the sampling in an adaptive fashion, so that the accuracy for measuring the quality of service parameters (delay, loss, jitter, throughput) is better if we know something about the traffic type and traffic parameters. Our study proposes and investigates a mechanism that can be set up to adaptively adjust the parameters of the sampling technique. Two realistic network topologies based on MPLS networks are setup to evaluate the proposed adaptive sampling scheme for monitoring and measuring network performance metrics. Compared with conventional sampling techniques (systematic and stratified sampling), simulation results are presented to illustrate that adaptive sampling provides the potential for better monitoring, control, and management of high-performance networks with higher accuracy.

Impact of Retransmission Mechanism on SIP Overload: Stability Condition and Overload Control

SIP (Session Initiation Protocol) has been widely adopted as a signaling protocol to establish, modify and terminate media sessions between end-users in the Internet. SIP introduces a retransmission mechanism to ensure the reliability of its real-time message delivery. However, retransmission can make server overload worse, leading to server crashes in SIP-based carrier networks (e.g. Skype). In order to study the impact of retransmission mechanism on SIP overload, in this paper, we create a discrete time fluid model to describe the queuing dynamics of an overloaded SIP server. Then we derive a sufficient stability condition that a SIP server can handle the overload effectively under the retransmission mechanism. Fluid model allows us to run fluid-based Matlab simulation directly to evaluate the overload performance. Event-driven OPNET simulation was also conducted to validate our fluid model. Our simulation results demonstrate that: (1) the sufficient stability bound is quite tight. The bound indicates that effective CPU utilization as low as 20% can still lead to an unstable system after a short period of demand burst or a temporary server slowdown. Resource over-provisioning is not a viable solution to the server crash problem; (2) by satisfying the stability condition, the initial queue size introduced by a transient overload can avoid a system crash. Such stability condition can help the operator to determine whether and when to activate overload control mechanism in case of heavy load. A simple overload control solution is also proposed.

De-Registration Based S-CSCF Load Balancing in IMS Core Network

IP Multimedia Subsystem (IMS) provides a common control layer over which the services can be easily accessed by fixed and mobile users. In order to guarantee performance to both service providers and end service subscribers, careful management and maintenance within the control layer must be fulfilled. In this paper, we propose a solution that utilizes a controlling entity to track the utilization of hotspot Session Initiation Protocol (SIP) server (i.e., S-CSCF) and to initiate de-registration procedure so that a sufficient number of subscribers at the overloaded S-CSCF are re-associated with the other S-CSCF(s) for subsequent session services. The proposed solution redistributes and smoothes the future traffic load over individual S-CSCF in an automatic manner. The goodness of the proposal is assessed through simulations using OPNET.

Modeling and simulation of SIP tandem server with finite buffer

Recent collapses of SIP servers (e.g., Skype outage) indicate that the built-in SIP overload control mechanism cannot mitigate overload effectively. We introduce our analytical approach by investigating an overloaded tandem server scenario. Our analytical model: (1) considers a general case that both arrival rate and service rate for signaling messages are generic random processes; (2) makes a detailed analysis of departure processes; (3) allows us to run fluid-based simulations to observe and analyze SIP system performance under some specific scenarios. This approach is much faster than event-driven simulation which needs to track thousands of retransmission timers for outstanding messages and may crash a simulator due to limited computing resources. Our numerical results help us reach a counterintuitive conclusion: A SIP system with a large buffer size may continuously exhibit overload and long queuing delay after experiencing a short period of demand burst or a temporary server slowdown. Small buffer size, on the other hand, can mitigate overload quickly by rejecting a large portion of the requests from a demand burst, and then resume normal operation after a short period of time. Furthermore, numerical results demonstrate that overload at a downstream server may propagate or migrate to its upstream servers and therefore cause widespread server crashes in a real SIP network.

Designing an ATM-based broadband network: an overview

As ATM technology moves from testbeds and trial systems to full-service deployment, the analysis and synthesis of ATM networks need more attention. This paper highlights the main areas where the multi-service network design differs from that of the present telecommunications network, and offers an overview of an integrated solution to the entire analysis and synthesis problem. More detailed analysis is provided in two main areas: the network-wide link sizing and the end-to-end performance at both the call and cell levels. The performance evaluation is based on a precise and computationally efficient method.

Modelling chaotic behaviour of SIP retransmission mechanism

The Session Initiation Protocol (SIP) retransmission mechanism can cause SIP network collapse with short-term overload. In this paper, we investigate with a fluid modelling approach the chaotic behaviour of the SIP retransmission mechanism in SIP networks. We capture the complex correlation structure in SIP systems through a detailed and novel queuing analysis. To dimension a buffer size which can avoid unnecessary message drop in a SIP server, we develop a sufficient condition for a stable SIP system analytically based on our fluid model. We also apply our fluid model to the simulation of a complex SIP system. We compare the simulation results achieved through our fluid models with those based on OPNET® event-driven approach to demonstrate the validity of our approach.

Modeling and design of a Session Initiation Protocol overload control algorithm

Recent collapses of Session Initiation Protocol (SIP) servers indicate that the built-in SIP overload control mechanism cannot mitigate overload effectively. In this paper, we propose a new SIP overload control algorithm by introducing a novel analytical approach to model the dynamic behavior of a SIP network where each server has a finite buffer. Three key breakthroughs of our modeling approach are the formulations of the message loss process, message retransmission process, and the complex departure process through detailed analysis. Our modeling results indicate that retransmissions triggered by the queuing delay are redundant, thus we propose a feedback control mechanism that regulates the retransmission message rate to mitigate the overload. We then demonstrate how to extend our analytical approach to the modeling of our overload control solution. Simulation based on this analytical model runs much faster than event-driven simulation, which needs to track thousands of retransmission timers for outstanding messages and may crash a simulator due to limited computation resources. Performance evaluation demonstrates that: (1) without the control algorithm applied, the overload at a downstream server may propagate to its upstream servers and cause widespread network failure; (2) in the case of short-term overload, our feedback control solution can mitigate the overload effectively without rejecting calls intentionally or reducing network utilization, thus avoiding the disadvantages of existing overload control solutions. In addition, compared with the pushback solution, our retransmission-based solution achieves a better trade-off between the speed to cancel the overload and the call rejection rate when an overload lasts a short period.

Controlling Retransmission Rate for Mitigating SIP Overload

With rapidly growing deployment, SIP has become a main signaling protocol for IP telephony and multimedia sessions in the Internet. SIP employs a retransmission mechanism to maintain its reliability. Recent server collapse due to emergency-induced call volume in carrier networks indicates that message retransmissions triggered by various SIP timers make the overload worse. The built-in overload control mechanism cannot handle overload conditions effectively. Since the retransmissions caused by the overload introduce more overhead rather than reliability, we suggest mitigating the overload by reducing the retransmission rate. We propose a novel algorithm to detect the potential overload at the downstream servers and control retransmission message rate from upstream servers to mitigate the overload at the downstream servers. We investigate two typical overload scenarios caused by demand burst and server slow down respectively. OPNET simulations demonstrate that (1) the proposed solution can help the overloaded downstream server to cancel its overload effectively after it resumes its normal operation status; (2) without the overload control algorithm applied, the overload at the downstream server may propagate or migrate to its upstream servers.