Citation Context

...stem model, an Aloha contention management model was assumed. Considering a more realistic CSMA model remains as future work. The approach will be to combine the CSMA Markov chain proposed by Bianchi =-=[29]-=- and the Markov chains proposed in this work together to have a bigger multi-dimensional Markov chain. Distributed algorithms The proposed medium access models do not rely on any central entity, so th...

Citation Context

...l unavailability events. Let us use the random variable Au to denote the inter-arrival time of those events. Au has a geometric distribution with probability pc. From the Renewal Theory point of view =-=[27]-=-, n = m − 1 is the number of renewals from 0 to Le where the process which is renewed is Au. The number of renewals can be found from Renewal Theory results [27]. However, as we have geometric distrib...

Citation Context

...ysis and packet level performance evaluation are required to investigate how a protocol, along with its parameter settings, behaves for delay-sensitive applications, such as multimedia communications =-=[9,10]-=-. Further, for cognitive radio networks, which consists of alternating operating and interruption periods, throughput analysis in the presence of saturated traffic is often insensitive to the period l...

Citation Context

...channel. Acknowledgments: This research project was supported by NSERC under Grant STPG365205. Les Cahiers du GERAD G–2014–97 1 1 Introduction Opportunistic spectrum access (OSA) communication models =-=[1]-=-, implemented by cognitive radios (CR), offer both the capacity to decrease the communication infrastructure expenses by using the vacant portions of the spectrum and to improve the quality of wireles...

Citation Context

...ngth of the remaining part of the packet for all users in service, which increases the state complexity. For the channel, a packet capture model with memory (e.g., a Markov chain channel fading model =-=[28]-=-) can be addressed as follows. When a node starts using a channel, the user has no knowledge of the state of the channel in the previous timeslot. Therefore, XT (similarly for Xu in the switching mode...

Citation Context

...n(N,Mc) links might thus simultaneously exist. Operations in the OSA network are time-slotted, and primary and secondary users are fully synchronized (common assumption to make the analysis tractable =-=[20, 25]-=-). In the beginning of each timeslot, there is a short quiet period during which the whole cognitive network stops operation to sense the channels. Primary users (PU) activity in any channel, includin...

Citation Context

...re [2–4]. Different protocols may differ in details such as the existence of a common control channel, contention management, number of radios required for the nodes and sensing algorithm (please see =-=[5, 6]-=- and references therein for a detailed survey of CR MAC protocols). However, what is common in most of the proposed protocols is the spectrum management and channel handover, which can be broadly clas...

Citation Context

...ot be easily extended to general models. In [19], two different switching models for sensor networks are discussed and the buffering model is not investigated. Instead of a common control channel, in =-=[20]-=-, a hopping model has been addressed and the focus is to evaluate the proposed proactive spectrum switching model. Instead of delay and throughput, [21] proposed effective bandwidth as the performance...

Citation Context

...s for which we derive the delay analysis under unsaturated traffic in later sections. More details on those protocols, as well as their throughput performance under saturated traffic, can be found in =-=[6, 7]-=-. It is assumed that there are N homogeneous nodes in an ad-hoc cognitive radio network and M OSA channels where one channel is the dedicated control channel and the remaining MC = M − 1 similar chann...

Citation Context

... full insights into the impact of different parameters on the network performance have not been provided. The target has also been specific networks; e.g., sensor networks, with specific requirements =-=[18, 19]-=-, such that the results can not be easily extended to general models. In [19], two different switching models for sensor networks are discussed and the buffering model is not investigated. Instead of ...

Citation Context

...re [2–4]. Different protocols may differ in details such as the existence of a common control channel, contention management, number of radios required for the nodes and sensing algorithm (please see =-=[5, 6]-=- and references therein for a detailed survey of CR MAC protocols). However, what is common in most of the proposed protocols is the spectrum management and channel handover, which can be broadly clas...

Citation Context

...ysis and packet level performance evaluation are required to investigate how a protocol, along with its parameter settings, behaves for delay-sensitive applications, such as multimedia communications =-=[9,10]-=-. Further, for cognitive radio networks, which consists of alternating operating and interruption periods, throughput analysis in the presence of saturated traffic is often insensitive to the period l...

Citation Context

... full insights into the impact of different parameters on the network performance have not been provided. The target has also been specific networks; e.g., sensor networks, with specific requirements =-=[18, 19]-=-, such that the results can not be easily extended to general models. In [19], two different switching models for sensor networks are discussed and the buffering model is not investigated. Instead of ...

Citation Context

...hroughput, [21] proposed effective bandwidth as the performance metric, but focus is on spectrum sensing and adaptive transmission and the discussion is provided for a single cognitive radio user. In =-=[11,22]-=-, we proposed a general M/G/1 queueing model [23] for a cognitive radio link where the server is subject to interruptions [24]. The channel (i.e., the server of the queue) is subject to interruptions ...

Citation Context

...igated. Instead of a common control channel, in [20], a hopping model has been addressed and the focus is to evaluate the proposed proactive spectrum switching model. Instead of delay and throughput, =-=[21]-=- proposed effective bandwidth as the performance metric, but focus is on spectrum sensing and adaptive transmission and the discussion is provided for a single cognitive radio user. In [11,22], we pro...

Citation Context

...performance metric, but focus is on spectrum sensing and adaptive transmission and the discussion is provided for a single cognitive radio user. In [11,22], we proposed a general M/G/1 queueing model =-=[23]-=- for a cognitive radio link where the server is subject to interruptions [24]. The channel (i.e., the server of the queue) is subject to interruptions because the CR node has to suspend its communicat...

Citation Context

...for the buffering MAC protocol. XR and XT are independent, so any moment of the service time X can easily be found from the moments of XR and XT . Discrete version of the Pollaczek?-Khinchine formula =-=[26]-=- can then be used to find the queue performance metrics of interest such as expected delay and delay variance (jitter). The exact distribution of XT and approximations for the distribution of XR are d...

Citation Context

...operating and interruption periods, throughput analysis in the presence of saturated traffic is often insensitive to the period lengths while packet delay for unsaturated traffic varies significantly =-=[11]-=-. It is therefore crucial to possess delay analysis tools to assess the performance of cognitive networks, which is the motivation for this paper. In the few papers available on delay analysis (e.g., ...

Citation Context

...n and the discussion is provided for a single cognitive radio user. In [11,22], we proposed a general M/G/1 queueing model [23] for a cognitive radio link where the server is subject to interruptions =-=[24]-=-. The channel (i.e., the server of the queue) is subject to interruptions because the CR node has to suspend its communication session for the period of spectrum handover (switching) or primary users’...