A distributed spectrum handoff MSRV protocol for the cognitive radio ad hoc networks Morteza Mehrnoush, Reza Fathi & Vahid T. Vakili

Wireless Networks The Journal of Mobile Communication, Computation and Information ISSN 1022-0038 Wireless Netw DOI 10.1007/s11276-017-1446-9

1 23

Your article is protected by copyright and all rights are held exclusively by Springer Science +Business Media New York. This e-offprint is for personal use only and shall not be selfarchived in electronic repositories. If you wish to self-archive your article, please use the accepted manuscript version for posting on your own website. You may further deposit the accepted manuscript version in any repository, provided it is only made publicly available 12 months after official publication or later and provided acknowledgement is given to the original source of publication and a link is inserted to the published article on Springer's website. The link must be accompanied by the following text: "The final publication is available at link.springer.com”.

1 23

Author's personal copy Wireless Netw DOI 10.1007/s11276-017-1446-9

A distributed spectrum handoff MSRV protocol for the cognitive radio ad hoc networks Morteza Mehrnoush1 • Reza Fathi2 • Vahid T. Vakili3

 Springer Science+Business Media New York 2017

Abstract Cognitive radio technology provides opportunistic wireless spectrums access for the secondary users (SUs) while primary users (PUs) are dormant. By emergence of a PU in the cognitive radio networks, SUs are required to vacant the channel by using spectrum handoff approaches to avoid any collision. Providing an efficient and flexible coordination protocol for spectrum handoff is a challenging step. In this paper, we propose a novel proactive spectrum handoff protocol called multiple-single rendezvous (MSRV) protocol which uses multiple rendezvous (MRV) and single rendezvous (SRV) coordination policy to provide a higher throughput and context adaptability. MRV coordination policy has the advantage of negotiating in different channels simultaneously and avoid single control channel congestion. On the other hand, in order to share the PUs’ channel usage history information between the SUs for predicting channels’ availability, it is necessary to utilize a SRV coordination scheme. The proposed MSRV protocol improves the average throughput of the SUs and decreases the average service time comparing to the other existing proactive spectrum handoff protocols. Moreover, MSRV protocol gives priority to the & Morteza Mehrnoush [email protected] Reza Fathi [email protected] Vahid T. Vakili [email protected] 1

School of Electrical Engineering and Computer Science, Washington State University, Pullman WA, USA

2

Department of Computer Science, University of Houston, Houston, TX, USA

3

School of Electrical Engineering, Iran University of Science and Technology, Tehran, Iran

handoff SUs which decreases service times. It is an crucial property for the delay sensitive applications. Our proposed protocol achieves 20% throughput and 13.7% average service time improvement in 1.25 and 10 (pkts/s) PUs’ traffic load compared to the SRV proactive spectrum handoff protocol proposed by Mehrnoush et al., respectively. Keywords Cognitive radio networks  Spectrum handoff  Rendezvous channel  Channel coordination protocol

1 Introduction In cognitive radio networks (CRNs), secondary users (SUs) coexist with the primary users (PUs) and other SUs [1]. Since the PUs has the priority in the CRNs, SUs are required to vacant the channel when PUs appear and try to continue their transmission in other channels [2]. It is known as the spectrum handoff in CRNs. There are two types of spectrum handoff in CRNs called reactive and proactive spectrum handoff [3]. In the reactive spectrum handoff, SUs sense the PUs’ appearance after the collision between the SU and PU, then they vacant the channel and try to continue their transmission in other channels [3, 4]. In the proactive spectrum handoff, SUs predict the PUs’ appearance by using channel usage history information or statistics parameters [5, 6]. In this approach SUs vacant the channel before appearance of the PUs in the channel. Although reactive spectrum handoff approaches do not require the channel usage history information which is an advantage, they need an extra delay of searching for idle channels and then performing spectrum handoff to achieve a consensus on the new channels [4]. In [7], SUs utilize a proactive spectrum handoff scheme based on the greedy channel selection (GCS) method

123

Author's personal copy Wireless Netw

for selecting a channel. In this scheme, the channel is selected based on channel usage history information and prediction of service time (time from starting the transmission of a packet until the end of it) for each channel. This scheme has a major issue of considering one pair of the SUs in the network while in a multi user network it causes inordinate collisions between the SUs [5]. In [8], a proactive spectrum handoff protocol based on time estimation is proposed which reduces the communication disruption and improves SUs throughput. This scheme is proposed for a network with only one pair of SUs which is a simplified case and is not useful in real networks. A spectrum handoff algorithm is proposed in [9] based on making a channel reservation table. When a PU appears in a channel, SUs can choose a channel from the channel reservation table and make a spectrum handoff. This table is updated periodically. A proactive spectrum handoff protocol using a common hopping coordination scheme is proposed in [10]. This protocol utilizes the channel usage statistics to derive some criteria for doing spectrum handoff. It decreases the interference between the SUs and PUs and increases the SUs’ throughput. The authors, in [11], consider the proposed proactive spectrum handoff criteria in [10] and introduced a common hopping coordination scheme in a multi users’ CRN. Authors in [5] extended the latter two methods (i.e., [10, 11]) and introduced a multiple rendezvous coordination protocol where more than one pair of SUs can contend simultaneously to access the channels. Another proactive spectrum handoff protocol is proposed in [12] where SUs can coordinate in a multi-user network to compete with other SUs for accessing the channel. The channel selection approach in this paper is based on a GCS scheme and results in a lower average delay and higher average throughput for the CRN. In [13], a spectrum handoff model based on a Hidden Markov model is proposed to decrease the sensing overhead in proactive spectrum handoff sensing and to analyze the status of the channel. In [14], a proactive unified spectrum handoff (PUSH) scheme in a CRN is proposed through an established route. In PUSH scheme, a handoff threshold is used to initiate the handoff proactively and this scheme require Common Control Channel (CCC). In [15], a proactive spectrum handoff protocol is proposed to improve throughput and service time of the SUs in the CRNs. Moreover, they introduced a theoretical analysis to evaluate the performance of the protocol. The dynamic spectrum availability of CRN brings up a challenge of designing an advanced protocol in order to increase network throughput and decrease the service time. In this paper, we propose a novel proactive spectrum handoff protocol for a multiuser CRN called multiple-single rendezvous (MSRV) protocol to answer this challenge. It adapts McMAC [16, 17] as a MRV protocol which uses a distributed algorithm for channel selection to provide

123

scalability and a higher throughput. McMAC does not use CCCs [18] where it could cause several problems such as control channel saturation, CCC coverage range (scalability), and control channel security in the CRNs. MRV protocol uses channel usage history information to select the best vacant channel and avoids collision between the PUs and SUs in order to lead to a higher throughput. Our proposed protocol also adapts common hopping [17, 19] as a SRV protocol in order to both share the channel usage history information and coordinate among the SUs. In this protocol, SUs select the channels for performing spectrum handoff based on a multi-user GCS scheme [15]. This paper uses the same network model as in [15]. In [15] we adapted a split phase coordination protocol (SRV spectrum handoff protocol) for the CRN where the SUs can coordinate for accessing the channel and share the channel usage history information every N time slots (N [ 10). So, the SUs that want to perform a spectrum handoff should wait N time slots to select a channel for transmission. Our proposed protocol in the present paper improves the performance of the network compared to the previous one in [15]. In summary, this study includes the following contributions: •

• •

•

Proposing a novel distributed proactive spectrum handoff protocol for a multi-user CRN which improves SUs’ average throughput. Alleviating the congestion issue of CCC by using rendezvous channel and distributed coordination policies. Giving priority to the handoff SUs in the CRN to decrease service time as a significant parameter for the delay sensitive services. Achieving throughput improvement and decreasing average service time at different PUs traffic load compared to the pertinent protocols.

The rest of this paper is organized as following. In Sect. 2, we illustrate the network model and our multi-user GCS channel selection policy. Section 3 presents the proposed MSRV proactive spectrum handoff protocol. In Sect. 4, we provide simulation results and compared them with other papers’ results. And finally we present our conclusion and future works in Sect. 5.

2 Network model and multi-user GCS channel selection policy In our considered CRN model, a SU performs spectrum handoff when a PU want to utilize its dedicated channel. Collision between the SUs and PUs is avoided because SUs can predict the emergence of the PUs by using the channel usage history information and go to the handoff state [15]. Then, those handoff SUs utilize the multi-user GCS scheme for channel selection to continue their interrupted

Author's personal copy Wireless Netw

transmission. In this section, we initially state the required network assumptions. Then, we present the channel selection criteria based on the multi-user GCS scheme for selecting the best channel for performing the spectrum handoff. 2.1 Network model assumption In our model, the channel follows an ON–OFF state. The dashed rectangles in the Fig. 1 show the packet transmission of the PUs as an ON state and other areas as an OFF state which illustrates that PUs do not have any packets for transmission. The PUs and SUs are a M/G/1 system and the packet arrival rates of both SUs and PUs follow the Poisson distribution process [3, 4]. Average arrival rates of the PUs and SUs are kp and ks , respectively. In order to make the model tractable, it is assumed that all the PUs and SUs utilize the licensed channels. We assume there are N SUs and M channels in the network. Since the power of the transmitted signal is higher than the received signal, the instantaneous collision detection and transmission is not possible for one radio wireless node. Therefore, in the proposed scheme it is assumed that each SU is provided with two radios [10, 11]. The first kind of radio is called the transmitting radio which is used to transmit data and control packets. The second kind of radio is called the scanning radio which is used to scan the channel and gather the channel usage history information. Moreover, the scanning radio should sense the selected channel to certify that the selected channel is not occupied by other SUs. 2.2 Multi-user GCS channel selection policy A multi-user GCS channel selection scheme [15] is an approach for channel selection in a CRN where all the SUs

predict achievable service time in each channel based on the channel usage history information. To perform a spectrum handoff, SUs select the best channel based on two criteria as: minimum service time and maximum vacant time. These criteria results in a lower service time and higher throughput. Figure 1 shows an example of a multiuser GCS scheme for one pair of SUs in a CRN. The transmitter SU selects the best channel based on two upcoming criteria. In order to make the manuscript self descriptive, we describe necessary parts from our previous research work [15]. Achieving a minimum service time is the first criteria in a multi-user CRN. For selecting the channel, SUs compare staying time in their channels with the time of changing to other channels:  Sk if Sk \Cj þ th STi ¼ ; Cj if Sk  Cj þ th ð1Þ j ¼ 1; :::; M&j 6¼ k where Sk denotes the staying time in the current channel of the SU (k is the current number of channel for i th SU), and Cj is the changing time of the channel which the SU can select (j is the number of changing channel for i th SU). STi is the minimum time of the selected channel by i th SU to continue the transmission. Let th denote the handoff delay for changing its channel to another channel. If the staying time is less than the changing time, the SU selects current channel and stays in its channel. If the changing time plus the handoff delay is less than the staying time, the SU decides to change its channel and use the second criterion for selecting the best channel. The second criterion is the maximum vacant time where SUs select the channel with maximum vacant time among the channels which have zero changing time: VTm ¼ sortðTm Þ;

m ¼ 1; :::; M

m6¼k

Fig. 1 The SU-1 starts the transmission to SU-2 in channel 1 in third time slot. After 11 time slots, a PU appears in channel 1 and hence, the SU should decide to either change the channel and perform a spectrum handoff or stay in the current channel and resume its unfinished transmission. In this example, the SU changes its channel and resumes the transmission on channel 2, because this channel has maximum vacant time. In time slot 18, a PU appears in the channel 2 and SU should decide to either stay in the current channel or change its channel and performs a spectrum handoff. In this example, SU decides to stay in its channel and continue the unfinished transmission after the PU finished its transmission

ð2Þ

where Tm is the vacant time of the channels which have zero changing time. It is the time duration from the instant SUs perform a spectrum handoff until PUs occupy the channel. In Eq. 2, the sort function arranges the vacant time of the SUs in a decreasing order. Vector VTm includes the vacant time of all channels which is arranged decreasingly. By considering these two criteria, SUs can select the best channels to continue the transmission.

3 The proposed MSRV protocol CRN is an opportunistic and a highly dynamic network where PUs and SUs become active and idle independently. SUs are trying to use PUs’ dedicated channels while PUs are idle. SUs should vacate PUs dedicated channels if they

123

Author's personal copy Wireless Netw

emerge in their dedicated channels. This dynamicity of channels’ availability requires an adaptable protocol to increase throughput in the CRN. Our proposed MSRV protocol tries to achieve this aim by using different policies at different situations based on the status of the SUs in the network. We describe SUs’ different status in subsection A. Then, we describe coordination policies in our MSRV protocol in detail in subsection B. In subsection C, we clarify how those policies work in our MSRV protocol. 3.1 Secondary users’ states Since SUs use different policies to make a decision based on their status, they reside in a predefined state during their presence in the network. Any SU is in one of the Idle, Negotiating, Communicating, or Handoff states. A SU which does not have any packet for transmission stays idle, so it is in the Idle state. When a SU generates packets for transmission, it goes to the Negotiation state and is ready for transmission. SUs are in the Communicating state when they are transmitting data. Those SUs which go to the Communicating state from an Idle state are called new communicating SUs. When a pair of SUs coordinate and occupy a channel for communication, they go to the Communicating state and start data transmission.They continue their transmission in that state till either they finish transmission and go back to the idle state, or a PU emerges in the channel. In the latter case, the SUs stop their transmission, vacate the channel for the PUs, and go to the Handoff state. Handoff SUs wait to compete at the next rendezvous channel to coordinate and access to a channel to resume their transmission. When an SU goes to the Handoff state, it decides either to stay in its current channel to resume its transmission later on or go to the Negotiating state to access another channel for continuing its’ transmission. For this aim, it utilizes the multi-user GCS policy and channel usage history information by considering Eq. (1). During the Negotiation state, SUs coordinate to select a channel for transmission as described in the next section. Corresponding to the state of the SUs, our MSRV protocol uses different policies to achieve a higher throughput and lower service time. 3.2 Coordination policies In order to coordinate for transmission, transmitter and receiver SUs need to meet at a channel and do handshaking. Our proposed MSRV protocol uses two different policies for coordination based on the SUs’ state. The first one is a MRV based policy which uses an adapted McMAC [16] protocol in CRN to make coordination among the new communicating SUs defined in Sect. 3.1. MRV uses a distributed algorithm to coordinate simultaneously among

123

multiple SUs. Any SU generates a hopping pseudo random sequence to follow by using their own MAC address as its seed and share this sequence (broadcast it) with all other SUs. Thus, all the SUs have different hopping sequences to follow and therefore they will try different channels at different time slots. When a SU generates a packet to transmit to another SU, it deviates from its default hopping sequence and follows its intended receiver SU’s hopping sequence. It is possible that more than one SU try to communicate with one of the receiver SUs which can cause a collision. So, they use carrier sense and collision avoidance policy to deal with this problem. They choose a window size and a random time within that window. In the specified random time, initially the transmitter SU senses the channel, then sends a RTS packet and senses the channel at the same time. If a collision occurs, SUs double the window size and reattempt later. If no collision occurred and the channel was vacant (it sends and receives an RTS/CTS successfully, i.e. a successful handshaking), they start transmission at the receiver’s hopping channel. For example, in Fig. 2, SUs 5 and 7 coordinate simultaneously with their intended receivers (SUs 6 and 8, respectively) by using the MRV policy in channels 2 and 6 at the second time slot, respectively; and then they immediately start to transmit their data. The second policy is SRV, which is used by the SUs in Handoff state. It enables the Handoff SUs to share the channel usage history information and be able to use the proactive spectrum handoff policy in their channel selection. They utilize the channel usage history information to estimate the PU’s appearance in the channel and select the best channel based on those mentioned criteria in Sect. 2.2. SUs use the OR operand for making the decision of the PUs appearance in the channel based on the shared channel usage history information. As a result, all the SUs make the same prediction about the PUs existence. Since all the SUs do the sensing, the probability of misdetection of a PU is very small. The SRV based policy utilizes a common hopping [19] protocol for coordination. All the SUs follow a common periodic hopping sequence for determining the rendezvous channel for coordination at each time slot. In Fig. 2, SRV rendezvous coordination channel (RCC) is illustrated at each time slot. In SRV, all Handoff SUs compete in one rendezvous coordination channel at a time slot. The time slot is divided into K 0 þ 1 mini slots. K 0 is the minimum number of the available SUs in the CRN and maximum possible mini slots in each time slot. Each Handoff SU follows a pseudo random sequence to handshake with its intended receiver SU in a mini slot corresponding to its sequence number. If the sequence number of some SUs are bigger than the maximum possible mini slot (K 0 þ 1), those SUs will not compete in the current rendezvous time slot. Since each SU is assigned to a

Author's personal copy Wireless Netw

Fig. 2 An example of utilizing MRV and SRV coordination policies in our MSRV protocol

different mini slot in each rendezvous time slot, it also provides fairness in the competition. For example, in Fig. 2, SUs 1 and 3 go to a Handoff state at the end of first time slot when PUs appear in the channels 1 and 5, respectively. In the second time slot, SUs 1 and 3 coordinate with their intended receivers (SUs 2 and 4, respectively) by using SRV policy in channel 3 and then they resume their transmission. 3.3 MSRV proactive spectrum handoff protocol In the CRNs, SUs try to occupy opportunistic channels dedicated to the PUs when they are dormant. Our MSRV protocol adapts itself to the dynamicity of the network with different PUs traffic loads. It adapts itself to this network dynamicity by letting the SUs use two different policies at different states. At each time slot, every SU follows the distributed coordination algorithm presented in Algorithm 1. At first, if the SU is a new communicating SU, it chooses to utilize MRV policy for coordination. The transmitter SU hops to its intended receiver SU’s channel by following its pseudorandom sequence except when it is in the same channel as the SRV rendezvous channel. If this exception happens, SU stops coordinating and waits for the next time slot. If not, it hops to the receiver’s channel and senses the channel for two mini slots. The first one is to make sure that the channel is vacant by the other SUs, and the second mini slot is for making sure that no Handoff SU had selected that channel for transmission from the previous coordination time slot. The latter mini slot sensing gives priority to the Handoff SUs. If the transmitter SU finds the channel vacant for two mini slots, it starts to do handshaking and if it is successful, it starts the transmission immediately. If the SU is a Handoff SU, it uses SRV policy for coordination. Any Handoff SU senses the first mini slot of

the SRV channel to make sure it is vacant. If it is vacant, it follows a pseudo-random sequence to generate its competing mini slot number in the SRV rendezvous channel. It goes to the SRV rendezvous channel for competition at its corresponding mini slot. Then, they start handshaking at their corresponding mini slot. If any pair of Handoff SUs can coordinate and handshake successfully, they go to their agreed channel in the next time slot. Then, sense the channel for a mini slot, perform handshaking, and start transmission. Consider that when a pair of Handoff SUs choose a channel from the list of available vacant channels, others will receive their successful handshaking RTS/CTS and will remove that channel from their list. Our proposed MSRV protocol provides priority to the Handoff SUs by requiring new Communicating SUs to sense their selected channel for two mini slots. Since Handoff SUs sense their selected channel just for one mini slot before starting their transmission, they will start sending data at the second mini slot. Consequently, if another pair of new communicating SUs have chosen that channel for coordination by using MRV, they find it occupied at the second mini slot and stay dormant. Indeed, the latter waiting one more mini slot avoids collision between the SUs, which are using SRV and SUs which are using MRV protocols at the same channel and time slot. This priority policy decreases service time of the applications which causes an interesting quality of service for the end users [20, 21]. When the PUs are highly dormant, their dedicated channels are most frequently vacant. So, new communicating SUs use MRV policy to coordinate concurrently and occupy those vacant channels for transmission. In this situation, few PUs emerge at their dedicated channels and hence few SUs face to vacate a channel and go to the Handoff state. Thus, the majority of the SUs use MRV policy for coordination. This allows many SUs to coordinate concurrently and hence it easily scales up. This

123

Author's personal copy Wireless Netw

scalability is a prominent characteristic of our MSRV protocol, which helps it to increase network’s usage throughput. On the other hand, if the network is highly used by the PUs, more spectrum Handoff happens, where more Handoff SUs use SRV policy to share the channel usage history information and coordinate. This more updated information helps SUs to choose the channels with higher precision and hence behave more intelligent to choose a channel. The difference between our previously proposed proactive SRV protocol in [15] and current MSRV proposed protocol is at the utilized coordination policies. In [15], all the SUs are using SRV for coordination. But, in the current MSRV protocol, only Handoff SUs uses SRV protocol while other SUs are using MRV coordination protocol. In [15], a time slot is divided into 2K 0 þ 1 mini slots instead of K 0 þ 1 mini slot compared to current proposed MSRV protocol. The SRV protocol in [15] used the first half mini slots to provide priority to the Handoff SUs, while the MSRV protocol does not need that because only Handoff SUs utilizes SRV protocol for coordination. MSRV protocol provides priority to the SUs by forcing users of the SRV protocol to sense their selected channel for mini slot before data transmission while users of the MRV protocol sense their selected channel for two mini slots. In a CRN there are two types of spectrum sensing errors called misdetection and false alarm [22]. In our protocol,

123

each SU performs spectrum sensing and shares this information with each other. So, SUs have a more accurate channel sensing information which decreases the probability of misdetection and false alarm. But if a collision occurs because of misdetection, SUs can detect it and vacate the channel because they all have an extra radio for sensing the channel. So, they leave the channel occupied by a PU and go to the Handoff state. And if a false alarm sensing error of a PU happens, the SUs will not transmit their packet on that channel. Another case is that if those issues happens between the SUs, the behaviour of the SUs is the same as the interaction between SU and PU which is already explained. In all cases, the average network throughput will decrease but it is negligible. In order to discuss the consequences of different sensing in the transmitter and receiver as a rare condition, we consider the following cases. If just the transmitter senses the channel busy, because of PUs appearance, it will stop transmitting. If just the receiver senses the PUs’ appearance, but not the transmitter, then the handshaking process (RTS/CTS) will fail and they will stop transmitting. In MRV and SRV policies in our MSRV protocol, SUs sense the channels, perform handshaking, and transmit the packet. If the sensing result is different in the transmitter and receiver, similar to the previous case it ends in an unsuccessful RTS/CTS handshaking. So, SUs try to handshake in the next time slot.

Author's personal copy Wireless Netw

4 Simulation results We presented simulation results in this part to evaluate the performance of our proposed MSRV protocol. It is compared to other relevant proactive spectrum handoff protocols in the literature. We tabulated the simulation parameters in Table 1. Figure 3 shows the average throughput of our proposed MSRV protocol, under different traffic loads of PUs. It shows that the performance of our MSRV protocol is better than the compared spectrum handoff protocols. Our MSRV protocol achieves 20% throughput improvement in 1.25 (pkts/s) PU traffic load compared to the SRV spectrum handoff protocol [15]. This improvement is achieved mainly due to the simultaneous coordination of the SUs in multiple channels. It provides a higher possibility to the SUs to compete with each other for accessing the channels. Moreover, in this protocol, SUs who want to do spectrum handoff can hop to a new coordinating channel in each time slot while in the SRV spectrum handoff protocol they should wait until the end of the data phase. So, SUs have a

Table 1 Simulation parameters Parameters

Values

The channel bit rate

R ¼ 1 (Mbps)

The length of the time slot

Ts ¼ 2 (ms) M ¼ 10

The number of SU

N ¼ 20

Data packet length of PU

105 (bits)

Data packet length of SU

6  104 (bits)

Packet generation rate of the SUs

500 (pkts/s)

16

550

14

500

Random Channel Selection Protocol

Average Service Time (time solts)

Secondary User’s Average Throughput (pkt/s)

Number of the channels

fewer possibility to compete for accessing the channel, and it decreases SUs’ average throughput. For example, let us assume in SRV spectrum handoff protocol the length of a data phase is 50 time slots [15]. If a SU finishes at any of the time slots (like in 20th), it has to wait until the end of the data phase (like 30 slots) to reach the next control phase to compete for accessing a channel to transmit any newly generated packets. MSRV protocol has 33.5% higher throughput compared to the probability based spectrum handoff protocol [5] in 1.25 (pkts/s) PU traffic load. In the following, we make an overhead comparison between the MSRV and SRV protocols. The control data transmission between the SUs for coordination and accessing the channel is the network overhead. There are two types of control data as overhead, the PUs’ channel usage history information sharing and handshaking of the SUs. Both overheads are higher in the MSRV in comparison with the SRV protocol. But the handshaking and channel usage history information sharing time is very short (a small fraction of a mini slot), so it does not decrease the network throughput. Frequent channel usage history information sharing helps to have a more accurate PUs’ appearance prediction in the channels and make a better decision. Further, more frequent handshaking provides more opportunities for the SUs to coordinate and access the channel for transmission, and consequently achieve a higher average throughput. In Fig. 4, we compared average service time of the SUs in our MSRV protocol and other existing spectrum handoff protocols. It shows that the average service time of our MSRV protocol is lower than the other compared protocols. Because SUs who want to start transmission for coordination with other SUs can coordinate in different channels at the same time slot, they have more possibilities

12 10 8 6 4 2 0 10−1

SRV Spectrum Handoff Protocol Probability Based Spectrum Handoff Protocol Random Channel Selection Protocol Proposed MSRV Spectrum Handoff Protocol

100

SRV Spectrum Handoff Protocol Probability Based Spectrum Handoff Protocol

450

Proposed MSRV Spectrum Handoff Protocol

400 350 300 250 200 150 100

101

Primary User Traffic (pkt/s)

Fig. 3 Average throughput of the SUs in our MSRV protocol in comparison with the other existing spectrum handoff protocols

50 10 −1

10 0

10 1

Primary User Traffic Load (pkt/s)

Fig. 4 Average service time of the SUs in our MSRV protocol in comparison with the other existing spectrum handoff protocols

123

Author's personal copy Wireless Netw

to compete and hence it takes fewer times for them to access a channel. Again, with the same reason, our MSRV protocol decreases average service time by 13.7% in PU traffic load of 10 (pkts/s) compared to the SRV spectrum handoff protocol. Our MSRV protocol has resulted in 38% lower average service time compared to the probability based spectrum handoff protocol [5] in PU traffic load of 10 (pkts/s). We consider there are 20 channels in the network and the packet transmission rate of the PUs are kp ¼ 5 (pkts/s) in Fig. 5, respectively. As the number of the SUs increases, the average throughput of the SUs decreases. This is because in this case there are more SUs to utilize the

Secondary Users’ Average Throughput (pkt/s)

16

5 Conclusion and future work

SRV Spectrum Handoff Protocol Probability Based Spectrum Handoff Protocol

14

Random Channel Selection Protocol Proposed MSRV Spectrum Handoff Protocol

12

10

8

6

4

2

5

10

15

20

25

30

35

40

Number of Secondary Users

Fig. 5 Average throughput of the SUs in our MSRV protocol in comparison with the other existing spectrum handoff protocol under different number of SUs

Secondary Users Average Throughput (pkt/s)

16 SRV Spectrum Handoff Protocol Random Channel Selection Protocol Probability Based Spectrum Handoff Protocol

14

channel and the number of the idle channels for packet transmission decreases. We show in Fig. 5 that the average throughput of the SUs in our MSRV protocol is better than the other proposed proactive spectrum handoff protocols. We assumed that there are 20 SUs in the network and the packet transmission rate of the PUs are kp ¼ 5 (pkts/s) in Fig. 6, respectively. Average throughput of the SUs increases by increasing the number of channels and converges to a flat throughput curve when the number of channels grows. It also reveals that the average throughput of the SUs in our MSRV protocol is much higher than the average throughput of the other existing proactive spectrum handoff protocols.

Proposed MSRV Spectrum Handoff Protocol

12

Cognitive radio technology allows the SUs to utilize the dedicated dormant channels of the PUs. Any CRN protocol faces a major challenge of collision avoidance between SUs and PUs as well as between the SUs themselves. We proposed a novel proactive spectrum handoff protocol called MSRV which uses two policies of MRV and SRV to mitigate this challenge. MRV coordination policy has the advantage of negotiating simultaneously in different channels. On the other hand, SRV coordination policy uses PUs’ channel usage history information to predict channels’ availability, in order to enable SUs to choose a channel with the highest vacant time for transmission. Our MSRV proposed protocol achieves 20% throughput improvement in 1:25 (pkts/s) PU traffic load than the SRV spectrum handoff protocol. The MSRV proposed protocol decreases average service time by 13.7% in PU traffic load of 10 (pkts/s) compared to the SRV spectrum handoff protocol. In the future work, we will propose an analytical analysis to probe this protocol more solidly. As another exploration, we will consider a backup channel reservation by the SUs which can provide an interesting feature for any critical service time application.

10

References

8

6

4

2

5

10

15

20

25

30

Number of Channels

Fig. 6 Average throughput of the SUs in our MSRV protocol in comparison with the other existing spectrum handoff protocol at different number of channels

123

1. Akyildiz, I. F., Lee, W.-Y., Vuran, M. C., & Mohanty, S. (2006). Next generation/dynamic spectrum access/cognitive radio wireless networks: A survey. Computer Networks, 50(13), 2127–2159. 2. De Domenico, A., Strinati, E. C., & Di Benedetto, M.-G. (2012). A survey on MAC strategies for cognitive radio networks. IEEE Communications Surveys and Tutorials, 14(1), 21–44. 3. Wang, L.-C., & Wang, C.-W. (2008). Spectrum handoff for cognitive radio networks: Reactive-sensing or proactive-sensins? In IEEE international performance, computing and communications conference (IPCCC) (pp. 343–348).

Author's personal copy Wireless Netw 4. Wang, C.-W., Wang, L.-C., & Adachi, F. (2010). Modeling and analysis for reactive-decision spectrum handoff in cognitive radio networks. In IEEE global telecommunications conference (GLOBECOM 2010) (pp. 1–6) IEEE. 5. Song, Y., & Xie, J. (2012). Prospect: A proactive spectrum handoff framework for cognitive radio ad hoc networks without common control channel. IEEE Transactions on Mobile Computing, 11(7), 1127–1139. 6. Xie, X., Yang, G., & Ma, B. (2011). Spectrum handoff decision algorithm with dynamic weights in cognitive radio networks. In Global mobile congress (GMC) (pp. 1–6) IEEE. 7. Wang, L.-C., Wang, C.-W., & Chang, C.-J. (2012). Modeling and analysis for spectrum handoffs in cognitive radio networks. IEEE Transactions on Mobile Computing, 11(9), 1499–1513. 8. Li, L., Shen, Y., Li, K., & Lin, K. (2011). TPSH: A novel spectrum handoff approach based on time estimation in dynamic spectrum networks. In IEEE international conference on computational science and engineering (CSE) (pp. 345–350) IEEE. 9. Liu, F., Xu, Y., Guo, X., Zhang, W., Zhang, D., & Li, C. (2013). A spectrum handoff strategy based on channel reservation for cognitive radio network. In International conference on intelligent system design and engineering applications (ISDEA) (pp. 179–182) IEEE. 10. Song, Y., & Xie, J. (2010). Proactive spectrum handoff in cognitive radio ad hoc networks based on common hopping coordination. In IEEE conference on computer communications workshops INFOCOM (pp. 1–2) IEEE. 11. Song, Y., & Xie, J. (2010). Common hopping based proactive spectrum handoff in cognitive radio ad hoc networks. In IEEE global telecommunications conference (GLOBECOM) (pp. 1–5) IEEE. 12. Mehrnoush, M., & Vakili, V. T. (2013). Proactive SRV spectrum handoff protocol based on gcs scheme in cognitive radio adhoc network. In International journal of power control signal and computation (IJPCSC) (pp. 1–8) IJPCSC. 13. Pham, C., Tran, N. H., Do, C. T., Moon, S. I., & Hong, C. S. (2014). Spectrum handoff model based on hidden markov model in cognitive radio networks. In International conference on information networking (ICOIN) (pp. 406–411) IEEE. 14. Nejatian, S., Syed-Yusof, S., Latiff, N., & Asadpour, V. (2014). Push: Proactive unified spectrum handoff in CR-MANETs. In International conference on advanced information networking and applications (AINA) (pp. 33–40) IEEE. 15. Mehrnoush, M., Fathi, R., & Vakili, V. T. (2015). Proactive spectrum handoff protocol for cognitive radio ad hoc network and analytical evaluation. IET Communications, 9(15), 1877–1884. 16. So, W., Walrand, J., & Mo, J., et al. (2007). McMAC: A parallel rendezvous multi-channel mac protocol. In IEEE wireless communications and networking conference (WCNC) (pp. 334–339) IEEE. 17. Mo, J., So, H.-S. W., & Walrand, J. (2008). Comparison of multichannel MAC protocols. IEEE Transactions on Mobile Computing, 7(1), 50–65. 18. Lo, B. F. (2011). A survey of common control channel design in cognitive radio networks. Physical Communication, 4(1), 26–39. 19. Tzamaloukas, A., & Garcia-Luna-Aceves, J. (2000). Channelhopping multiple access. In IEEE international conference on communications (ICC) (Vol. 1, pp. 415–419) IEEE. 20. Goel, A., Abeni, L., Snow, J., Krasic, C., & Walpole, J. (2002). Supporting time-sensitive applications on general-purpose operating systems. In Proceedings of the 5rd symposium on operating systems design and implementation (OSDI’02) (pp. 165–180).

21. Fathi, R., Salehi, M. A., & Leiss, E. L. (2015) User-friendly and secure architecture (UFSA) for authentication of cloud services. In 8th international conference on cloud computing (CLOUD) (pp. 516–523). 22. Zhang, H., Jiang, C., Mao, X., & Chen, H. H. (2016). Interference-limited resource optimization in cognitive femtocells with fairness and imperfect spectrum sensing. IEEE Transactions on Vehicular Technology, 65(3), 1761–1771.

Morteza Mehrnoush (S’09) received his M.S. degree in electrical engineering from the Iran University of Science and Technology, Tehran, Iran, in Feb. 2013 and the Ph.D. degree in electrical engineering from the Washington State University, Pullman, WA, USA, in Aug. 2016. He is currently working as a post-doctoral researcher at University of Washington, Seattle, WA, USA. His current research interests include wireless communication, channel coding, signal processing for communications, turbo equalization, 2-D magnetic recording, and cognitive radio networks. Reza Fathi received his Master of Science in Software Engineering from Iran University of Science and Technology in 2011. He is currently a Ph.D. student at University of Houston. He is interested in design and analysis of randomized distributed algorithms for large scale graphs and big data analytics, cloud computing and security.

Vahid T. Vakili received the B.S. degree from Sharif University of Technology, Tehran, Iran, in 1970, the M.S. degree from the University of Manchester, Manchester, UK, in 1973, and the Ph.D. degree from the University of Bradford, Bradford, UK, in 1977, all in electrical engineering. In 1985, he joined the Department of Electrical Engineering, Iran University of Science and Technology, Tehran. He has served as the Head of the Communications Engineering Department and as the Head of postgraduate studies. His research interests are in the areas of mobile cellular systems, interference cancellation for CDMA systems, and space-time processing and coding.

123