S. Lirio Castellanos-Lopez

Joint Call and Packet Level Performance Analysis of CAC Strategies for VoIP Traffic in Wireless

Call admission control (CAC) functionality is a critical requirement for guarantee the desired quality of service (QoS) for voice calls in IP-based wireless networks. In this paper, three different CAC strategies for voice over IP (VoIP) traffic over wireless access networks with packet buffering are mathematically analyzed through a joint call and packet level discrete time teletraffic model. The admission criterion (i.e., admission threshold) of new sessions of these strategies is based on either the total number of sessions, the number of active sessions (i.e., sessions in talk-spurt periods), or the number of packets queued in the buffer. Admission threshold and buffer size can be both controlled for QoS provisioning in terms of session blocking and packet dropping probabilities. Arrivals and completions of VoIP sessions, on/off activity detection, and periodic and constant length packet generation (of active sessions) of individual VoIP sessions are modeled. The developed teletraffic analysis allows to evaluate the performance of the CAC strategies in terms of the most relevant QoS metrics of VoIP traffic at both call and packet level (i.e., session blocking and packet dropping probabilities and packet delay). Finally, the maximum traffic attained by the different studied-CAC strategies (while QoS provisioning is guaranteed) is obtained.

Erlang capacity in coordinated cognitive radio networks with stringent-delay applications

Due to the unpredictable nature of channel availability, supporting the quality of service (QoS) of stringent delay sensitive traffic in cognitive radio networks (CRNs) is very challenging. Stringent delay sensitive calls in CRNs are susceptible to forced termination due to the preemptive resource occupancy priority of primary users. To enable more efficient usage of the spectrum while improving the QoS experienced by secondary users (SUs), dynamic spectrum leasing (and coordinated cognitive radio) of resources has been previously proposed in the literature. In this paper, the Erlang capacity in coordinated cognitive radio networks with real-time hard delay constraint traffic as function of the rented resources is analytically calculated. For the adequate and fair performance comparison, call admission with fractional channel reservation to prioritize ongoing secondary calls over new ones is considered. From numerical results, it is observed that in CRNs exists a critical utilization factor of the primary resources from which it is not longer possible to guarantee the required QoS of SUs and, therefore, delay sensitive services cannot be even supported in cognitive radio networks. Thus, spectrum leasing can be essential for CRN operators to provide the QoS demanded by delay sensitive services. Then, the cost per capacity Erlang as function of both the utilization factor of the primary resources and the maximum allowed number of simultaneously rented channels is evaluated.

Performance analysis of coordinated cognitive radio networks under fixed-rate traffic with hard delay constraints

Due to the unpredictable nature of channel availability, carrying delay-sensitive traffic in cognitive radio networks (CRNs) is very challenging. Spectrum leasing of radio resources has been proposed in the so called coordinated CRNs to improve the quality of service (QoS) experienced by secondary users (SUs). In this paper, the performance of coordinated CRNs under fixed-rate with hard-delay-constraints traffic is analyzed. For the adequate and fair performance comparison, call admission control strategies with fractional channel reservation to prioritize ongoing secondary calls over new ones are considered. Maximum Erlang capacity is obtained by optimizing the number of reserved channels. Numerical results reveal that system performance strongly depends on the value of the mean secondary service time relative to the mean primary service time. Additionally, numerical results show that, in CRNs without spectrum leasing, there exists a critical utilization factor of the primary resources from which it is not longer possible to guarantee the required QoS of SUs and, therefore, services with hard delay constraints cannot be even supported in CRNs. Thus, spectrum leasing can be essential for CRN operators to provide the QoS demanded by fixed-rate applications with hard delay constraints. Finally, the cost per capacity Erlang as function of both the utilization factor of the primary resources and the maximum allowed number of simultaneously rented channels is evaluated.

Simplified joint call and packet level teletraffic analysis of CAC strategies for VoIP traffic in wireless networks

In this paper, a simplified joint call and packet level teletraffic analysis method for performance evaluation of call admission control (CAC) strategies for voice over IP (VoIP) traffic over wireless access networks is proposed. The admission criterion of new sessions of the considered CAC strategies is based on either the total number of sessions or the number of active sessions. Arrivals and completions of VoIP sessions, on/off activity detection, and periodic and constant length packet generation of individual VoIP active sessions are modeled. Packet buffering, adaptive modulation coding, and proportional fair sharing scheduling are also considered. Our proposed analysis is based on the quantized queue and cross-level linked decomposition mechanisms that we introduce in this paper to drastically simplify the mathematical and computational procedures with no major impact on the accuracy of the results. The developed analysis allows evaluating the performance of the CAC strategies in terms of the most relevant QoS metrics of VoIP traffic at both call and packet level and, at the same time, mathematical and computational complexity are reduced with no major impact on the accuracy of numerical results.

VoIP Erlang Capacity in Coordinated Cognitive Radio Networks

Guaranteeing QoS is challenging in cognitive radio networks due to the dynamically changing network resources, especially for stringent delay sensitive traffic. Dynamic spectrum leasing (and coordinated cognitive radio) of primary network resources has been previously proposed in the literature to improve QoS of secondary users. In this paper, the Erlang capacity in coordinated cognitive radio networks with VoIP traffic is investigated. Employing time-scale decomposition technique and conventional continuous and discrete time Markov chain tools, a joint connection-level and packet-level analytical model is developed. For the adequate and fair performance comparison, a call admission strategy with fractional admission threshold and fractional buffering is considered. An algorithm to determine the optimal values (i.e., the ones that maximize Erlang capacity) of the admission threshold and buffer size is proposed. For given QoS requirements at session and packet level (i.e., VoIP session blocking, packet dropping, and packet delay), the Erlang capacity as function of the maximum allowable number of simultaneously rented resources is evaluated. Additionally, the fraction of time that spectrum leasing is needed is calculated.

Erlang Capacity Performance Evaluation of Spectrum Adaptation Strategies in Cognitive Radio Networks

Interruption of ongoing calls of secondary users (SUs) due to the arrival of primary users (PUs) is one of the most relevant factors that degrade both quality of service and system throughput in cognitive radio networks (CRNs). In this paper, the Erlang capacity achieved by different resource management mechanisms previously proposed in the literature to mitigate the effects of call interruptions is evaluated and compared when they are employed separately in cognitive radio networks (CRNs) with heterogeneous traffic. This evaluation is done to identify the most relevant mechanisms to improve the performance of these networks. Also, strategies that jointly employ different resource management techniques and spectrum adaptation to exploit the synergy of the different mechanisms are proposed and evaluated. The aim of the proposed spectrum allocation strategy is to take advantage of the flexibility and delay tolerance features of elastic traffic. Additionally, a mathematical model for the performance analysis of the proposed strategies in CRNs with heterogeneous traffic is developed. Numerical results demonstrate that call buffering and adaptive spectrum mechanisms in heterogeneous traffic CRNs can significantly improve overall system performance by efficiently exploiting the elasticity of delay-tolerant traffic.

Performance evaluation of cognitive radio systems with Coxian distributed channel holding time in the primary network

In this paper, a general teletraffic analysis to evaluate the performance of cognitive radio networks with arbitrary order Coxian distributed channel holding time in the primary network is developed. The relevance of the considered Coxian model relies on its universality property, mathematical tractability, reduced dimensionality, and the fact that it generalizes a large class of useful previous employed distributions in the literature to characterize the channel holding time (i.e., exponential, Erlang, hyper-exponential, and hypo-exponential). Additionally, recently, there have been major advances in fitting Coxian distributions. Our numerical results for 2nd order Coxian model show negligible differences compared with those obtained when the log-normal distribution is considered. (Empirical evidence supports the log-normal assumption for the primary channel holding time.) Thus, the analytical model presented in this paper represents a step towards a general and mathematically tractable framework for the performance evaluation of cognitive radio networks under more realistic assumptions.

Superimposed training for conventional and code-aided timing recovery in turbo-coded systems

Initial timing error estimation is a critical issue for the convergence of iterative code-aided (CA) timing recovery algorithms in turbo-coded systems, especially at the low signal-to-noise ratio scenarios that turbo codes are designed to work. The proposed solutions in the literature involve additional frame synchronization processing or sending a short synchronization sequence. In this paper, the use of superimposed training (ST) to initialize iterative code-aided timing recovery algorithms in turbo code systems is proposed. Simulation results show that timing recovery based on ST can achieve a performance close to that of perfect synchronization for both CA and conventional timing recovery schemes. Furthermore, ST used for timing recovery can also be exploited for frame synchronization.