Hybrid automatic repeat request assisted cognitive radios

Abstract

It is widely known that the Cognitive Radio (CR) paradigm has the potential of improving the exploitation of the earmarked but momentarily unoccupied spectrum, which is exclusively allocated to the primary users (PUs) based on the conventional fixed spectrum allocation policy. The CR systems first have to sense, whether the PU’s band is unoccupied and then dynamically access it. Naturally, CR systems suffer from the same propagation impairments as the traditional wireless communication systems, such as interference, fading and noise, which affect both the reliability and the attainable data rate. In order for the CR system to achieve both reliable data transmission as well as a high throughput and low delay, we propose novel CR-aided Hybrid Automatic Repeat Request (HARQ) protocols, which intrinsically amalgamate the CR functions with HARQ protocols and study their performance. Both perfect and imperfect spectrum sensing are considered.

Specifically, we propose the cognitive stop-and-wait-HARQ (CSW-HARQ), cognitive Go-Back-N-HARQ (CGBN-HARQ) as well as the cognitive selective-repeat (CSR-HARQ) schemes and study their throughput and delay both by analysis and simulation. To protect the PUs legal rights, we model their activity of occupying a primary radio (PR) channel as a two-state Markov chain consisting of ‘ON’ and ‘OFF’ states. In order to use the PR channel, the CR system first senses the presence of the PUs and once the PR channel is found to be free (i.e., in the OFF state), the CR system transmits its data packets relying on the principles of SW-HARQ, GBN-HARQ and SR-HARQ. Otherwise, the CR system continues sensing the channel until finding a free one. Naturally, the PR channel may be sensed erroneously, which results either in false alarm or in misdetection. Therefore, the channel may be modelled by a two-state Markov chain, provided that sensing is ideal, or by a four-state Markov chain, if sensing is non-ideal. Here, the four states are determined by the actual state of the PR channel and the state sensed by the CR system. We analyse both the throughput and delay of CR systems relying on different HARQ schemes. We invoke a pair of analytical approaches, namely the probability based approach and the Discrete Time Markov Chain (DTMC) based approach. Closed-form expressions are derived for the throughput, average packet delay and the end-to-end packet delay. Furthermore, for the end-to-end packet delay, we derive both the probability distribution and the average end-to-end packet delay. In the DTMCbased approach, we propose a state generation algorithm for eliminating the illegitimate states, which helps reduce both the dimensionality of the related state transition matrices and the associated computational complexity. All the equations obtained by analysis are validated by numerical simulations.

Our performance results reveal that both the achievable throughput and delay of the CSWHARQ, CGBN-HARQ and the CSR-HARQ schemes are substantially affected by the activity of the PUs, by the reliability of the PR channels, by the unreliable sensing decisions and by the number of packets transmitted per time-slot (TS). Specifically, when the probability of the PR channel being busy is high and/or its reliability is relatively low, the throughput attained by these HARQ schemes becomes relatively low and their packet delay increases. Furthermore, for the CGBN-HARQ and CSR-HARQ, our results show that when the propagation environment is time-variant, the number of packets transmitted within a TS should be adapted accordingly, in order to attain the highest throughput and the shortest average transmission delay.

More information

Identifiers

ORCID for Lieliang Yang: orcid.org/0000-0002-2032-9327

Catalogue record

Export record