Prioritized channel allocationbased dynamic spectrum access in cognitive radio sensor networks without spectrum handoff
 JongHong Park^{1} and
 JongMoon Chung^{1}Email authorView ORCID ID profile
https://doi.org/10.1186/s1363801607644
© The Author(s) 2016
Received: 16 June 2016
Accepted: 31 October 2016
Published: 21 November 2016
Abstract
In this article, a dynamic spectrum access (DSA) method is proposed to perform dynamic prioritized channel allocation in cognitive radio sensor networks (CRSNs) that do not support spectrum handoff. The proposed CRSN supports primary users (PUs) and two types of secondary users (SUs), where one type of SU data traffic has a higher priority. The proposed prioritized channel allocation (PCA) scheme was designed to provide adaptive spectrum access control for CRSNs supporting SU prioritized data types. Compared to the existing DSA schemes, SNO and DSAC2, the proposed PCA scheme shows an advantage in throughput and fairness in supporting SU prioritized traffic.
Keywords
1 Introduction
Cognitive radio (CR) sensor networks (CRSNs) have been proposed to improve the channel utilization of wireless networks suffering from serious shortage of radio spectrum. In CRSNs, licensed users use assigned channels as primary users (PUs). On the other hand, each sensor node which does not have a license to access a channel is a secondary user (SU), where SUs attempt to access a channel only if it is not used by a PU. In recent studies, dynamic spectrum access (DSA) strategies for CRSNs have been considered [1–6]. DSA channel assignment schemes can be categorized as centralized, distributed, and decentralized [2]. Centralized schemes have a central node that collects channel and link information from the sensor nodes [7–9]. The collected information is used to execute the channel assignment algorithm to determine the channel assignments for all of the links in the wireless sensor network. In distributed schemes, each node calculates the channel environment and condition of its local links and selects the appropriate channel considering local information [10–12]. A decentralized DSA can be operated in clusterbased wireless sensor networks, where a cluster head performs the intracluster channel assignment in a centralized form and computes the intercluster channel assignment in a distributed manner [13, 14]. The distributed and decentralized DSA schemes have an advantage in not requiring an additional central control system. However, since the nodes in the CRSN only use local information, distributed and decentralized DSA schemes may not result in optimal results [2]. In addition, a majority of existing research on DSA for CRSNs assume that SUs are equipped with spectrum handoff (SHO) functionality. As SHO requires additional operations (e.g., scanning for open channels, transmitterreceiver handshake, etc.), CR devices supporting SHO will consume additional energy and also a CR channel controller is needed [15–17]. Due to these reasons, there are CRSNs that do not have SHO functionality. In this article, a prioritized channel allocation (PCA) scheme is proposed to enhance the performance of CRSNs that do not support SHO functionality in a centralized DSA scheme. The proposed CRSN scheme supports PUs and two types of CR SUs, where one type of SU data traffic has a higher priority over the other type. Priority distinction in SU data packets was found needed in many CRSN applications where control/management messages/data and alarm information requires higher priority support from the CRSN. Therefore, the proposed PCA scheme was designed to provide adaptive spectrum access control for CRSNs supporting prioritized SU data types.
In the proposed PCA scheme, the entire subchannel was designed to be divided for each class of SUs without SHO functionality. After the allocation of PUs, each SU is allocated. By separating sections of subchannels for SUs and assigning subchannels from the highest number, the transmission success rate and connection sustaining rate of SUs with higher priority can be optimized.
The state transitions of the proposed PCA scheme are modeled as a multidimensional Markov chain with three state variables. Based on the state transition modeling, the blocking probability (BP), forced termination probability (FTP), and the call completion rate (CCR) are analyzed for the SUs with and without priority service support in the CRSN.
The rest of this paper is organized as follows. In Section 2, the system model and the proposed PCA scheme for centralized CRSNs are presented. In Section 3, the analytical models based on Markov chains with respect to the proposed PCA scheme are provided. In Section 4, the performance measures including the BP, FTP, and CCR for both SU _{1} and SU _{2} are derived. In Section 5, the numerical and simulation results under the proposed PCA scheme are presented. Finally, Section 6 gives the concluding remarks.
2 System model
In the proposed PCA scheme, the subchannels are divided into two sections. One section is for SU _{1}, where α (0<α<MN) subchannels are assigned among MN subchannels. The other section is for SU _{2} users, where up to MN−α subchannels can be assigned. In the proposed scheme, channel allocation of PUs are assigned in an ascending sequential order (i.e., 1, 2, …,M) and the SU _{1}s scan and use the CR subchannels in a descending sequential order from MN to MN−α+1 (i.e., MN,MN−1,…,MN−α+1). In addition, the SU _{2}s scan and use the CR subchannels in a descending sequential order from MN−α to 1 (i.e., MN−α,MN−α−1,…, 1).
The CRSN that applies the proposed PCA scheme is fully connected, which means that all SUs observe the same channel status. It is assumed for simplicity that a PU call requires one channel, whereas a SU call requires one subchannel. In addition, a common control channel is assumed to exist for coordination among the SU _{1}s and SU _{2}s, and perfect sensing is assumed to detect the PUs and SUs activity. The bandwidth requirement of SU _{1} and SU _{2} is assumed to be identical, and there is no specific geolocation requirement of SU _{1}s and SU _{2}s.
where T _{1} and T _{2} are the CCR of SU _{1} and SU _{2} calls, respectively. The CCR of SU _{1} and SU _{2} are the mean number of calls which is determined by the BP and FTP of SU _{1} or SU _{2}, described in detail in Section 4. FI is a measure of how fair the channel resources are shared among SU _{1}s and SU _{2}s [21]. The lower bound of FI, ε, is a minimum required value of fairness and FI is upper bounded by 1. A higher FI value is an indicator of a fairer network.
3 Markov chain model
3.1 Case 11: from state (i,j,k) to other states when k=0

If iN+j=MN, the ongoing N packets of SU _{1} will be forced to terminate by the arrival of a PU. Then, a state transition (i,j,0)→(i + 1,j − N,0) will occur with rate ϕ(i + 1,j − N,0) λ _{ p }.

If (M − 1)N < iN + j < MN, and j = p ^{′} N + q ^{′} (0 ≤ p ^{′} ≤ M,0 ≤ q ^{′} < N), the arrival of a PU will drop the ongoing q ^{′} packet(s) of SU _{1}. Accordingly, state transition (i,j,0)→(i + 1,p ^{′} N,0) will occur with rate ϕ(i + 1,p ^{′} N,0) λ _{ p }.

If iN + j ≤ (M − 1)N, the arrival of a PU will be assigned one channel without any SU _{1} user being forced to terminate. In this case, state transition (i,j,0)→(i + 1,j,0) will occur with rate ϕ(i + 1,j,0) λ _{ p }.
2) PU completes service In this case, state transition (i,j,0)→(i − 1,j,0) will occur with rate ϕ(i − 1,j,0) i μ _{ p }. 3) SU _{1} Requests Service In case of k = 0, SU _{1} can request for service and access an unoccupied subchannel if iN+j<M N. In this case, state transition (i,j,0)→(i,j + 1,0) will occur with rate ϕ(i,j + 1,0) λ _{ s1}.
4) SU _{1} completes service In this case, state transition (i,j,0)→(i,j − 1,0) will occur with rate ϕ(i,j − 1,0) j μ _{ s1}.
5) SU _{2} requests service SU _{2} can request for service and access an unoccupied subchannel when both iN + j < MN and i < M − p. In this case, state transition (i,j,0)→(i,j,1) will occur with rate ϕ(i,j,1) λ _{ s2}.
3.2 Case 12: from other states to state (i,j,k) when k=0

If iN + j = MN and p ^{′} = p, the following states can transfer to state (i,j,0) with rate ϕ(i − 1,j ^{′},k ^{′}) λ _{ p } based on (i − 1,j ^{′},k ^{′}) →(i,j,0) where j ^{′}+k ^{′} ≤ N,0 ≤ j ^{′} ≤ q, and 0 ≤ k ^{′} ≤ N − q. In this case, j ^{′} SU _{1}s and k ^{′} SU _{2}s suffer from forced termination when a PU accesses the channel.

If iN + j = MN and p ^{′} < p, the following states can transfer to state (i,j,0) with rate ϕ(i − 1,j + n,0) λ _{ p } based on (i − 1,j + n,0)→(i,j,0) where 0 ≤ n ≤ N. In this case, only n SU _{1}(s) suffer(s) from forced termination when a PU accesses the channel.

If iN + j < MN, the arriving PU will be assigned one channel without any SU _{1} and SU _{2} being forced to terminate. In this case, state transition with rate ϕ(i − 1,j,0) λ _{ p } will result in (i − 1,j,0)→(i,j,0).
2) PU completes service In this event, there is only one possibility for which the system will transit to state (i,j,k) with rate ϕ(i + 1,j,0)(i + 1) μ _{ p }, when iN + j ≤ (M − 1)N, which is (i + 1,j,0)→(i,j,0).
3) SU _{1} requests service In case of k=0, this event results in a transition (i,j − 1,0)→(i,j,0) with rate ϕ(i,j − 1,0) λ _{ s1}.
4) SU _{1} completes service When iN + j < MN, state transition (i,j + 1,0)→(i,j,0) will occur with rate ϕ(i,j + 1,k) (j + 1)μ _{ s1}.
5) SU _{2} completes service SU _{2} can request service and access an unoccupied subchannel for the case when both iN + j < MN and i < M − p are satisfied. In this case, the state transition (i,j,1)→(i,j,0) will occur with rate ϕ(i,j,1) μ _{ s2}.
3.3 Case 21: from state (i,j,k) to other states when k>0

If iN + α + k = MN and i < M − p − 1, the arrival of a PU will results in a drop of the ongoing N packets of SU _{2} without any SU _{1} being forced to terminate. Then, there is only one state to which the system will transfer to (based on rate ϕ(i + 1,j,k − N) λ _{ p }), which is (i,j,k)→(i + 1,j,k − N).

If iN + α + k = MN and i = M − p − 1, the arrival of a PU will results in a drop of all ongoing packets of SU _{2}. In addition, if j > pn, then (j − pn) packets of SU _{1} will be forced to terminate. Otherwise, there is no packet of SU _{1} to be dropped. Therefore, the state transition (i,j,k)→(i + 1, min[pN,j],0) will occur with rate ϕ(i + 1, min[pN,j],0) λ _{ p }.

If (M − 1)N < iN + j < MN and i < M − p − 1, the arrival of a PU will drop the ongoing packet(s) of SU _{2} without any SU _{1} being forced to terminate. Then, the state transition (i,j,k)→(i + 1,j,(M − p − i − 1)N − q) will occur with rate ϕ(i + 1,j,(M − p − i − 1)N − q) λ _{ p }.

If (M − 1)N < iN + j < MN and i = M − p − 1, the arrival of a PU will drop all ongoing packets of SU _{2}. In addition, if j > pn, then (j − pn) packets of SU _{1} will be forced to terminate. Otherwise (i.e., j ≤ pn), no packet of SU _{1} needs to be dropped. Therefore, the state transition (i,j,k)→(i + 1, min[pN,j],0) will occur with rate ϕ(i + 1, min[pN,j],0) λ _{ p }.

If iN + j ≤ (M − 1)N, the arrival of a PU will be assigned one channel without any SU _{1} and SU _{2} being forced to terminate. In this case, state transition (i,j,k)→(i + 1,j,k) will occur with rate ϕ(i + 1,j,0) λ _{ p }.
2) PU completes service In this case, state transition (i,j,k)→(i − 1,j,k) will occur with rate ϕ(i − 1,j,k) i μ _{ p }.
3) SU _{1} requests service SU _{1} can request for service and access an unoccupied subchannel when both iN + j + k < MN and j < α are satisfied. In this case, state transition (i,j,k)→(i,j + 1,k) will occur with rate ϕ(i,j + 1,k) λ _{ s1}.
4) SU _{1} completes service In this case, state transition (i,j,k)→(i,j − 1,k) will occur with rate ϕ(i,j − 1,k) j μ _{ s1}.
5) SU _{2} requests service SU _{2} can request for service and access an unoccupied subchannel when iN + α + k < MN. In this case, state transition (i,j,k)→(i,j,k + 1) will occur with rate ϕ(i,j,k + 1) λ _{ s2}.
6) SU _{2} completes service In this case, state transition (i,j,k)→(i,j,k − 1) will occur with rate ϕ(i,j,k − 1) k μ _{ s2}.
3.4 Case 22: from other states to state (i,j,k) when k>0

If iN + α + k = MN, with rate ϕ(i − 1,j,k + n) λ _{ p }, the states based on 0 ≤ n ≤ N can transfer to state (i,j,0) in the form of (i − 1,j,k + n)→(i,j,k). In this case, only n SU _{2}(s) suffer(s) from forced termination when a PU accesses the channel.

If iN + α + k < MN, the arrival of a PU will be assigned one channel without any SU _{1} and SU _{2} being forced to terminate. In this case, state transition (i − 1,j,k)→(i,j,k) will occur with rate ϕ(i − 1,j,k) λ _{ p }.
2) PU completes service In this event, there is only one possibility for which the system will transit to state (i,j,k) with rate ϕ(i + 1,j,0) (i + 1)μ _{ p }, when iN + α + k ≤ (M − 1)N, which is (i + 1,j,k)→(i,j,k).
3) SU _{1} requests service This event results in a transition with rate ϕ(i,j − 1,k) λ _{ s1} from (i,j − 1,k)→(i,j,k),
4) SU _{1} completes service In this case, the state transition (i,j + 1,k)→(i,j,k) will occur with rate ϕ(i,j + 1,k) (j + 1)μ _{ s1}.
5) SU _{2} requests service In this case, state transition (i,j,k − 1)→(i,j,k) will occur with rate ϕ(i,j,k − 1) λ _{ s2}.
6) SU _{2} completes service SU _{2} can request service and access an unoccupied subchannel based on the condition of iN + j<MN. In this case, state transition (i,j,k + 1)→(i,j,k) will occur with rate ϕ(i,j,k + 1) (k + 1)μ _{ s2}.
4 Performance analysis
4.1 Blocking probability (BP)
4.2 Forced termination probability (FTP)
4.3 Call completion rate (CCR)
5 Numerical results
The proposed PCA scheme is compared to two other DSA schemes supporting prioritized sensor data traffic in CRSNs, which are the sequential number ordering (SNO) scheme [9], and the DSA scheme with no buffer for centralized CRSN (DSAC2) [7]. Since the computational complexity is determined by the number of licensed channels (M) and the number of subchannels (N), the proposed PCA scheme and the compared schemes (i.e., SNO and DSAC2) in this paper have the same computational complexity [24]. The computational complexity of the SNO [9] and DSAC2 [7] schemes are identical to O(M ^{2} N ^{2}) and that of the proposed PCA scheme is O((MN−α)^{2}+α ^{2}), which results in the same complexity of O(M ^{2} N ^{2}) as SNO and DSAC2. Simulation is conducted for the case where all schemes do not use SHO while supporting two class prioritized traffic based on the settings of M=3, N=6, μ _{ p }=2, λ _{ s1}=0.2, λ _{ s2}=0.4, μ _{ s1}= μ _{ s2}=1, and ε=0.9.
6 Conclusions
In this article, an optimal CR channel allocation scheme that supports two types of prioritized sensor data types in CRSNs without SHO is proposed. The proposed PCA scheme was designed to provide optimal spectrum access control for CRSNs supporting heterogeneous SU data types. The performance analysis results show that when compared to the SNO scheme [9] and the DSAC2 [7], the proposed PCA scheme can provide a significant advantage in FTP and CCR while satisfying the FI constraint in supporting SU prioritized traffic. For future work, improved modeling of DSA for prioritized traffic considering dynamic spectrum fragmentation/defragmentation and improved spatial channel reuse based on the geolocations of SUs will be conducted.
Declarations
Acknowledgments
This work was supported by the ICT R&D program of MSIP/IITP. [R0126161009, Development of Smart Mediator for Mashup Service and Information Sharing among ICBMS Platform].
Competing interests
The authors declare that they have no competing interests.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
Authors’ Affiliations
References
 M Song, C Xin, Y Zhao, X Cheng, Dynamic spectrum access: from cognitive radio to network radio. IEEE Wirel. Commun. 19(1), 23–29 (2012).View ArticleGoogle Scholar
 E Ahmed, A Gani, S Abolfazli, LJ Yao, SU Kahn, Channel assignment algorithms in cognitive radio networks: taxomony, open issues, and challenges. IEEE Commun. Surv. Tutor. 18(1), 795–823 (2016).View ArticleGoogle Scholar
 A Ahmad, S Ahmad, MH Rehmani, NU Hassan, A survey on radio resource allocation in cognitive radio sensor networks. IEEE Commun. Surv. Tutor. 17(2), 888–917 (2015).View ArticleGoogle Scholar
 AK Mir, A Akram, E Ahmed, J Qadir, A Baig, in Unified channel assignment for unicast and broadcast traffic in cognitive radio networks. Proceeding of the 12th IEEE International Workshop on Wireless Local Networks (WLN) (Clearwater, FL, USA, 2012), pp. 799–806.Google Scholar
 M Talal, E Ahmed, J Qadir, A Baig, in Quantifying the multiple cognitive radio interfaces advantage. Proceedings of 6th International Conference on Signal Processing and Communication Systems (Gold Coast, Australia, 2012), pp. 12–14.Google Scholar
 A Bhattacharya, R Ghosh, K Sinha, D Datta, BP Sinha, Multimedia channel allocation in cognitive radio networks using FDMFDMA and OFDMFDMA. CoRR, vol. abs/1603, 03938 (2016). http://arxiv.org/abs/1603.03938.
 VK Tumuluru, P Wang, D Niyato, W Song, Performance analysis of cognitive radio spectrum access with prioritized traffic. IEEE Trans. Veh. Technol. 61(4), 1895–1906 (2012).View ArticleGoogle Scholar
 TMC Chu, H Phan, HJ Zepernick, Dynamic spectrum access for cognitive radio network with prioritized traffics. IEEE Commun. Lett. 18(7), 1218–1221 (2014).View ArticleGoogle Scholar
 JM Chung, CH Kim, JH Park, J Shin, D Kim, Optimal channel allocation for cognitive radios. IET Electron Lett. 46(11), 802–803 (2010).View ArticleGoogle Scholar
 E Ahmed, J Qadir, A Baig, Highthroughput transmissionqualityaware broadcast routing in cognitive radio networks. Wirel. Netw. 21(4), 1193–1210 (2015).View ArticleGoogle Scholar
 A Bhattacharya, R Ghosh, K Sinha, D Datta, BP Sinha, Noncontiguous channel allocation for multimedia communication in cognitive radio networks. IEEE Trans. Cog. Commun. Netw. 1(4), 420–434 (2015).View ArticleGoogle Scholar
 JH Park, JM Chung, Enhanced channel access estimation based adaptive control of distributed cognitive radio networks. KSII Trans. Internet Inf. Syst. 10(3), 1333–1343 (2016).MathSciNetGoogle Scholar
 F Saffre, R Tateson, R GhaneaHercock, Reliable sensor networks using decentralised channel selection. Comput. Netw. 46(5), 651–663 (2004).View ArticleGoogle Scholar
 JH Park, Y Nam, JM Chung, Sustainability enhancement multihop clustering in cognitive radio sensor networks. Int J Distrib. Sens. Netw. 11(10 574340), 1–8 (2015).Google Scholar
 A Ahmad, S Ahmad, MH Rehmani, NU Hassan, A survey on radio resource allocation in cognitive radio sensor networks. IEEE Commun. Surv. Tutor. 17(2), 888–917 (2015).View ArticleGoogle Scholar
 G GuR, S Bayhan, F Alagoz, Cognitive femtocell networks: an overlay architecture for localized dynamic spectrum access. IEEE Wirel. Commun. 17(4), 62–70 (2010).View ArticleGoogle Scholar
 AO Bicen, EB Pehlivanoglu, S Galmes, OB Akan, Dedicated radio utilization for spectrum handoff and efficiency in cognitive radio networks. IEEE Trans. Wirel. Commun. 14(9), 5251–5259 (2015).View ArticleGoogle Scholar
 VK Tumuluru, P Wang, D Niyato, A novel spectrumscheduling scheme for multichannel cognitive radio network. IEEE Trans. Veh. Technol. 60(4), 1849–1858 (2011).View ArticleGoogle Scholar
 Q Liu, S Zhou, GB Giannakis, Queuing analysis with adaptive modulation and coding over wireless links: Crosslayer analysis and design. IEEE Trans. Wirel. Commun. 40(3), 1142–1153 (2005).Google Scholar
 H Su, X Zhang, Crosslayerbased opportunistic MAC protocols for QoS provisionings over cognitive radio wireless networks. IEEE J. Sel. Areas Commun. 26(1), 118–129 (2008).View ArticleGoogle Scholar
 H Jiang, P Wang, W Zhuang, A distributed channel access scheme with guaranteed priority and enhanced fairness. IEEE Trans. Wirel. Commun. 6(6), 2114–2125 (2007).View ArticleGoogle Scholar
 D Bertsekas, R Gallager, Data networks, 2nd Ed. (PrenticeHall, USA, 1992). ISBN: 0132009161.MATHGoogle Scholar
 A Alshamrani, X Shen, LL Xie, QoS provisioning for heterogeneous services in cooperative cognitive radio networks. IEEE J. Sel. Area Commun. 29(4), 819–830 (2011).View ArticleGoogle Scholar
 D Kivanc, G Li, H Liu, Computationally efficient bandwidth allocation and power control for OFDMA. IEEE Trans. Wirel. Commun. 2(6), 1150–1158 (2003).View ArticleGoogle Scholar