Sie sind auf Seite 1von 10

Distributed Coordinated Spectrum Sharing MAC Protocol for Cognitive Radio

Hao Nan*, Tae-In Hyon**, Sang-Jo Yoo*. *Graduate School of Information Technology & Telecommunication, Inha University, 253 Yonghyun-dong, Nam-gu, Incheon 402-751, Korea {haonan1102@gmail.com , sjyoo@inha.ac.kr } http://multinet.inha.ac.kr ** Samsung Advanced Institute of Technology Communication & Network Lab San 14, Nongseo-Ri, Giheung-Eup, Yongin-Si, Gyeonggi-Do, Korea taein.hyon@samsung.com.
Abstractrecently, the CR (cognitive radio) technology is gathering more and more attention because it has the capacity to deal with the scarce of the precious spectrum resource. Within the domain of CR technology, channel management of CR is of utmost importance due to its key role in the performance enhancement of the transmission and the minimum interference to the primary users as well. An 802.11 WLAN based ad-hoc protocol using the cognitive radio has been proposed in this paper. It provides the detection and protection for incumbent systems around the communication pair by separating the spectrum into the data channels and common control channel. By adding the available channel list into the RTS and CTS, the communication pair can know which data sub channels are available (i.e., no incumbent signal). We proposed an ENNI (exchanging of neighbor nodes information) mechanism to deal with the hidden incumbent device problem. The simulation results show that by using our protocol the hidden incumbent device problem (HIDP) can be solved successfully. Key words Cognitive Radio, HIDP, CST, and CHRPT.

I. INTRODUCTION The rapid increasing of the wireless applications and productions result in the situation that there is no redundant frequency band can be allocated for such amazing trend due to the allocation manner of the current spectrum. Traditionally, spectrum allocation is fixed for each user for the wireless communication. As the needs for wireless communication application increasing, there is little spectrum can be allocated. Besides, the fixed channel allocations result in the low efficiency of the spectrum utilization. Under the observation of the FCC (Federal Communication Commission) [1], it has been proved that the most of the time, most of the allocated spectrum remains unutilized. Thus, how to take the advantage of the spectrum hole (spectrum assigned to the primary uses that not being used at specific time and location) is getting more and more attentions. Through making it possible for an unlicensed user to access a spectrum hole, the spectrum utilization can be enhanced significantly. Therefore, cognitive radio technology [2] is proposed under the supervision of IEEE 802.22 working group [3] to solve such
*This work was supported in part by Samsung Advanced Institute of Technology. *This research was supported by the MIC (Ministry of Information and Communication), Korea, under the ITRC (Information Technology Research Center) support program supervised by the IITA (Institute of Information Technology Assessment) (IITA-2006-C1090-0603-0019)

problem by taking advantage of the spectrum hole for unlicensed spectrum users in the way that co-exist with legacy users, through sensing and reporting. Nodes can adaptively select an arbitrary spectrum for the communication if it is not being used by the incumbent system. An 802.11 WLAN based ad-hoc protocol DCA (dynamic channel allocation) using the cognitive radio technology [4] has been proposed to deal with the multiple channel management issues through adding the available data channel list into the RTS (ready to send) and CTS (clear to send) before the communication begins. Upon receiving the RTS from sender, through comparison the available channel list between source and destination. Receiver chooses one channel for packet transmission. Then both of the sender and receiver switch to the reserved channel to begin the packet exchanging and set the corresponding NAV (network allocation vector) in the RTS and CTS packet that transmitted on common control channel to indicate the usage of the channel. DCA assume that each node equipped with two transceivers, one always monitor the common channel, which can avoid the hidden node problem consequently. Any node who wants to begin the transmission has to check the common control channel to make sure that the channel it wants to use is a clear channel. If there is no channel available then, node who wants to transfer packets has to wait for an idle channel through observation of common control and wait for a random backoff time for the fairness of contention for the network access, similar to the access contention mechanism of IEEE 802.11 DCF (distributed coordination function). However, due to the different interference range (range of the signals effect), there may appear some incumbent systems that the communication can not aware of during the channel selection period. As a result, the transmission will definitely cause some influence to licensed users on such channel that disobeys the original principle of CR: the implementation of the CR must not based on the cost of the primary users performance. Here, both of the incumbent device and primary users are the name of licensed users. In this paper, we propose a new MAC protocol implemented in the wireless ad hoc network that enables hosts to choose the suitable data channels that does not have any impact on licensed users through the information exchanging of the channel status

240 1-4244-0663-3/07/$20.00 2007 IEEE

of nodes. The core mechanism of our proposed protocol is the introduction of the time slot mechanism that used for the detection of incumbent devices around the communication pair that may not be detected directly by themselves. We add a time slot listening mechanism during the RTS and CTS exchanging period to report the licensed users by neighbor nodes of communication pair. The rest is organized as follows. Section 2 shows some background and identify the hidden incumbent device problem; Section 3 presents our proposed protocol in detail. Section 4 presents the results of our simulations. Section 5 discusses some issues related to our protocol that needs to be improved and the conclusion as well. II.

NAV (network allocation vector), which is included in both of the RTS and CTS. Therefore, through the RTS-CTS-DATA-ACK procedure of the DCF, the hidden node problem of single channel as that in Fig 1 can be avoided.

Fig. 1. Hidden node and expose node problems in single channel environment However, when it comes to the multi-channel environment, the DCF mechanism can not work well as mentioned by Jungmin So in [5]. In the multi-channel environment, nodes listen to different channels. It is very difficult to avoid the hidden node problem by using the RTS/CTS mechanism of IEEE 802.11 DCF. Fig 2 shows such problem. In Fig 2, node A wants to transmit a packet for node B, so it sends the RTS to node B on common control channel to notify neighbor nodes to avoid choosing the currently using channel as a transmission channel by other nodes who want to begin the message exchanging by using the NAV. After receiving the RTS, B selects channel 2 as the data channel and send the CTS to A to reserve the channel 2 as the transmission channel within the transmission range. There is no collision right now. However, when node B sent the CTS to A, node C was receiving on channel 3, it did not hear the CTS from B. Thus, after completing the transmission on channel 3, node C will assume that channel 2 is clear, and if it wants to begin the data transmission with node B, it is possible that node C just choose channel 2 as the data channel which will cause the collision to devices working on the channel 2. Therefore, [5] proposed MMAC (multi channel MAC) protocol to handle the multi channel hidden terminal problem mentioned above. In MMAC, periodically transmitted beacons divide time into beacon intervals. It uses the ATIM (ad hoc traffic indication messages) window (similar mechanism as that in IEEE802.11 PSM) to negotiate the channel during each beacon interval, and use the ATIM-RES (ATIM reservation) and ATIM-ACK to notify the nodes in the vicinity of sender and receiver. Nodes who can hear such message should update their PCL (preferable channel list)

BACKGROUND AND HIDDEN INCUMBENT DEVICE PROBLEM

In this part, we begin to describe a new hidden node problem called the hidden incumbent devices problem (HIDP). First, we give some related works that have already been proposed to deal with the multiple channels management issues. It is obviously that the multiple channel usage can enhance the total throughput significantly due to the reason that transmission can be processed on different channels which will not cause the collision and there is no interference to each other. However, under most situations, because of some economical reasons, node in the network can not install as many transceivers as we want which is not sufficient to monitor all the channels that being used. This will incur some problems that reduce the throughput of the whole network. Consequently, many protocols have already been proposed to handle multiple channel efficiency issues. Among them, how to solve the hidden node problem as described in the some protocols plays a key role in reducing the degradation to the performance of the system to minimum level. Traditionally IEEE 802.11 has a mechanism DCF to deal with the hidden node problem by using the RTS/CTS exchanging mechanism. In Fig 1, the first part shows the hidden node problem: node B is exchanging message with node C, node D can not sense signal of B by carrier sensing, it believes that the channel is free and begin to transmit packet to C which will cause the collision at node C. Thus D is the hidden node to B. while the expose node problem being showed in the second part of Fig 1, B is now exchanging message with node A. C can overhear such message, it believes that the channel is busy and has to wait until the end of the transmission of B. Actually, node C is able to communication with node D, thus node C is the expose node to node B. In the DCF mechanism of IEEE 802.11, it provides a RTS/CTS exchanging method to avoid the problem mentioned above. If a node wants to send packets to the destination node, it has to send the RTS which includes the time duration that other node should defer to the receiver. While at the destination side, upon receiving the RTS, it sends the CTS to notify the neighbor node around to defer their transmission as well in order to avoid the possible collision. The structure that records the time duration called

Fig. 2. Multi-channel hidden terminal problem

241

during the ATIM period. Thus, multi channel hidden terminal problem could be solved. During the ATIM window, all nodes must listen to the common control channel which is predefined, after the ATIM window; both sender and receiver switch to the channel reserved during the ATIM time. Besides, MMAC protocol provides a mechanism that the common control channel can be changed into the data channel which is useful for the situation that lack of the data channel. Then both sender and receiver switch to the agreed channel to begin the RTS -CTS-DATA-ACK procedure to start their data transmission. Through setting each channels statuses in the HIGH, MID, and LOW. Node who wants to transmit the packet can know the best channel that could be used. Thus, the overall system performance can be enhanced. Also, it only needs one transceiver for running while avoid the multi-channel hidden terminal problem which is economical compared with other multi-channel protocols with multiple transceivers. Dual busy tone multiple accesses [6] has been proposed to deal with the hidden node problems through switch the ON and OFF status of tone on the common control channel, it separate the frequency band into one common control channel and a sub data channel. By monitoring the status of the BUSY TONE of the common control channel, it can avoid hidden node problem. However, this method can avoid the hidden node problem only in the single data channel when it comes to the multiple channels, performance is not good. Soft channel reservation [7] is proposed by Nasipuri to select a channel that used for the last successful transmission as idle channel to begin the packet transmission. In short, the main concept in the multiple channel management is as being showed in Fig 3. However, there is a substantial issue which can degrade the performance of multiple channel networks. We select a protocol that based on the 802.11 mechanism named as dynamic channel assignment (DCA) [4] to describe the common problem that exist in the current multi-channel protocols. It maintains one dedicated channel (common channel) for exchanging the control messages and other channels for data transmission. Each node in the network equipped with two transceivers. Therefore, it can monitor the control channel and the data channel simultaneously. RTS/CTS packets are exchanged on the common control channel, data packets are transmitted on the data channel. In RTS packet, the sender includes a list of available channels. On receiving the RTS, the receiver determines one optimum channel for data transmission and includes the selected channel in the CTS packet. Further, in order to avoid the hidden node problem, NAV (network allocation vector) has also been used. Both the RTS and CTS contain the NAV to inform its neighbor node about the channel usage. After the sender received the CTS, source and destination nodes switch to the agreed channel and DATA and ACK packets are exchanged. As that in Fig 4, suppose there are four channels available at source node side, the available channel means that there are no other devices using such channel to transmit currently within

Fig. 3. General channel selection procedure of multiple channels protocols. ACL (available channel list) the sensing range of corresponding node (SC1, SC2, SC3, SC4 in this case). Source node uses the common control channel to send its current channel status information included in the RTS to the destination if it wants to begin its transmission. Then, after receiving the RTS, the destination compares the available channels of source node with its own available channel list.

Fig. 4. Channel selection procedure of original protocol Here, we just assume that each node has the capacity to sense environment around and use S for source node and D for the destination node. If there are available channels to D from the channel list of S, D chooses the one that is suitable for both direction. For instance, in Fig 4, after receiving the RTS from S, D chooses the SC1 as the data channel, and then D and S switch their transceivers to the agreed data channel and begin to exchange the DATA and ACK. Thus, it seems that we can choose a relatively good subset of channel which will not cause any interference to incumbent devices around, however, due to the different interference range, and the different sensing range of the communication pair as well, there exist some incumbent systems that can not be detected by the communication pair. Let us consider about the phenomenon as being showed in Fig 5. Around the neighborhood of source and the destination, there may appear some incumbent devices within the transmission range of the current communication pair which can not be sensed by the source and destination. Thus, not only the data packet from the sender, but also the ACK of the receiver can influence the incumbent systems around. This will definitely cause the interference to the performance of those incumbent

242

devices. Thus those sub-channels that incumbent systems occupied must be removed from the available channel list. In Fig 5, there are two incumbent systems inside sources transmission range. However, the source can not detect incumbent system signal because their radio waves can not reach the source. It is possible that after the exchanging of the RTS and CTS, destination side just choose the sub-channel 1 as the data channel which definitely will cause some interference to the incumbent system that working on the sub-channel 1 in this case. Within the sensing range of D, there are two sub data channels (SC3, SC4) taken up by incumbent devices, the primary user device on sub data channel 3(SC3) can be sensed by the D, therefore, the available channel list of D does not include SC3, however, in case of the SC4, it will be ignored by the communication pair with very high probability due to its small interference range. It will be considered as the background noise by the destination. We name this kind of the problem as the hidden incumbent device problem (HIDP). Next section, we will illustrate how our proposed protocol solves the HIDP.

always listen to the common control channel, while the other one locates on the data channels that can switch among different data channels. There are two main contributions of our proposed protocol; i) neighbor node reporting and ii) the mechanism of how to report. In addition to the functions that DCA (dynamic channel allocation) [4] has, our protocol has the mechanism that supervises hidden incumbent devices through the reporting of the unavailable data-channels occupied by the incumbent users that maybe affected by the currently communication pair around the neighborhood. We use the time slots mechanism to detect the hidden incumbent devices as showed in section 3.1. Besides, there is no synchronization needed in our proposed protocol compared with the MMAC while solving the hidden incumbent device problem successfully. Our protocol can solve the hidden multi-channel terminal problem [5] as well. We will show this in section 3.2 3.1. Entire Procedure of Proposed Protocol Our protocol is based on the DCA mechanism. Thus, In terms of the network access, we use the similar mechanism as that of IEEE 802.11 DCF mode. After source sends the RTS which includes the available channel list of itself (based on the source nodes channel status table, RTS includes M available channels) to the destination, sender arranges the corresponding time slots in the CHRPT-S (channel report slots to source marked by sources neighbors) period. Neighbor nodes who hear the RTS just compare the ASCL (available sub channel list) in the RTS with their own CST (Channel Status Table containing the available sub-channel list and the channel status:SINR), and find the channel that does not exist in its ACT (available channel table that contains the comparison result of the available channel list between sender and receiver), and then reply with the pulse in the specified time slot during CHRPT-S. Since ASCL (available sub channel list) has M SCs (sub channels), we have M CHRPT-S slots as well. Before sending the RTS, source node check its available sub channel list in the CST (channel status table that we will explain later), and arrange the numbers of time slot according to the amount of its currently free channels around. Then, receiver transmits a short pulse during the slot time if the channel corresponding to the slot is used by incumbent systems around themselves. For instance, if the first sub-channel of the sources available channel list has being occupied by the incumbent systems it will reply in the first time slot of CHRPT-S. Fig 6 shows the detection of the pulse. Because this kind of the pulse reports is generated by the neighbor node, the distance is very short; consequently, the propagation delay can be ignored. Therefore, there is no need to use the synchronization. We use the specific timeslot to represent the corresponding sub-data channel. Moreover, because the time duration of time slot of 802.11 is greater than that of the pulse. Thus, any pulse

Fig. 5. Hidden incumbent device problem III.


PROPOSED PROTOCOL

We describe the proposed protocol in this section. Our protocol is based on the assumptions below. There are N available data channels for communication. There always exists one common control channel for the control information exchanging. The entire sub data channels are non-overlapping channels. The spectrum is separated into several frequency bands that there is no over lapping between each other. Each node in the network has the capacity to sensing its environment around. And has the knowledge of which subset of data channels should be added to its own available channel list. It can transmit and receive on any of these channels. Each host is equipped with two transceivers; one is

243

appears with the range of specific time slot will be regarded as the appearance of the incumbent device on that sub data channel.

3.2 NAV (Network Allocation Vector) The NAV of our proposed protocol is similar to the IEEE802.11 DCF (distributed coordination function). Before transmission, source node and destination should exchange the RTS (request to send) and CTS (clear to send) on the common control channel to preserve the data channel. Other nodes in the network who hear this kind of control message must delay their transmission. NAV records the delay period. Thus, nodes who can hear the RTS and CTS are able to know how long they can access the channel again which will reduce the collision greatly. In our proposed protocol, we separate the NAV into CCNAV (common control channel NAV) used to indicate when other nodes can access the common control channel again and DCNAV (data channel NAV) for indicating the deferring time period that other node want to access the same sub data channel. Here, the CCNAVu has been used for the indication of the updated NAV of common control channel for reducing the access delay of the common control channel. This can avoid the hidden control message problem as well. Due to the same reason as that of the single channel hidden node problem, hidden problem can happen on the control message as well. As we mentioned before in the single channel environment in Fig.1, node B want to transmit RTS to C, without the CCNAV, node D can not know the communication between B and C, D will assume that the common control channel is available for C, if it want to initiate the transmission with C, there are some collisions occur which will cause the loss of control packets. Through adding the CCNAV, each node can know the usage of common control channel around while not only relay on the CS (carrier sensing) mechanism. The whole system throughput can also be promoted. Next we give a set of calculation method to the values will be used in the protocol. Eq.1 shows the calculation of CCNAV. SIFS (Short inter-frame spacing) is the shortest waiting time for medium access for arranging CHRPT; sending 2nd RTS (the updated RTS), CHREQ, CTS, DATA, and ACK. DIFS is the distributed inter-frame spacing used for first RTS transmission. Eq. 2 shows the UIFS which is the Update inter-frame spacing defined in this paper used for the indication of time period to transmit the RTSu. We use the PT as the time duration of pulse in the following equations.
CCNAV = UIFS + CHREQ + SIFS + M PT + SIFS + CTS

Fig. 6. Pulse detection Consequently, the corresponding sub data channel will be eliminated from the available channel list which can reduce the interference to the primary user greatly. Other method such as require the neighbor node to send its own channel situation will cause the collision. Because the pulse does not need to be decoded, sender can know which sub channel has already been used. If the source detects any pulse signal during CHRPT-S period, it will send the RTSu (RTS updated) which includes the ASCL (available sub-channel list), CCNAVu (CCNAV updated), DCNAV (data channel NAV). The ASCL of RTSu contains m sub channels based on information collected during the CHRPT-S (m M). If no pulse were detected, there is no need to transmit the RTSu packet to the destination node. At receiver side, D-Nx means the neighbor nodes x of destination, after receiving the first RTS, the destination node has to wait for certain time UIFS (update inter-frame spacing used to indicate the updating time) to get the possible RTSu (if there is no RTSu, it will deal with the first RTS). After the UIFS time, the receiver sends the CHREQ (channel status request) to its neighbors, which includes destinations ASCL containing k sub channels among the listed sub-channels (k m M) of source. Then neighbors of destination who hear the CHREQ reply with pulse during CHRPT-D (channel report slots to destination marked by destinations neighbors) according to their CSTs. Finally destination chooses one optimal data channel in the left ASCL of its own after the CHRPT-D period. Next step is to add the DCNAV which is the same as that in the RTS and selected channel into the CTS and send it back to the source node. If only j timeslots left in the ASCL that means that only j sub channels are available currently without interference to the incumbent systems around. Neighbor nodes who can hear the RTS and CTS must defer their transmission according to the NAV that included in them. In this protocol we use two transceivers; one is always listen to the common channel in order to update the CST and the other transceiver located on the sub data channels which can switch among them dynamically used for the packet transmission. After receiving the CTS at the receiver side, both source and destination nodes switch to the reserved channel to begin the transmission.

(1) (2)

UIFS = M PT + 2 SIFS +

Eq. 3 shows the length of the CCNAVu. If there is no pulse can be detected during the CHREQ, M equals to m.

244

Fig. 7. Whole procedure of our proposed protocol


CCNAVu = SIFS + CHREQ + SIFS + m PT + SIFS + CTS

(3)
Channel ID Sub data channel 1 Sub data channel 2 Sub data channel 3 Sub data channel N

TABLE I CHANNEL STATUS TABLE SINR 3


4 7 1

Eq. 4 shows the duration of CCNAV-D (CCNAV of destination side), which is included in the second RTS (RTSu), here, we use m PT instead of k PT , because if we can not terminate the CCNAV at the same time with that of source. Neighbors of the destination will get higher common control channel access probability than neighbors of the source.
CCNAV D = SIFS + m PT + SIFS + CTS

status

Available
Occupied Occupied Available

(4)
CST which will be included in the RTSu and being sent to the destination side. Destination node also keeps the same kind of table, upon receiving the RTSu, it compare its own one with the received one and retains the same available channel and record them in the ACT (available channel table). Then require the answer of it neighbor if they have any negative reply about the ACT currently. Finally, destination determines the data channel from the ACT according to the SINR of each channel. Then include it in the response CTS. It also includes DCNAV and begins to send to the source node. After the source node receives the CTS, both source and destination nodes switch to the agreed channel to begin the packet transmission. While in the implicit sensing mode, the sensing the available channels and updating of the CST depends on the neighbors message overhearing. Node does not have to sense the channel on its own initiative. They only have to listen to the message sent by its neighbor. Through the overhearing, each node can get the information of which channel has already been used by its neighbor, and consequently construct the CST and update it as well. Besides there exist the expiration mechanism in out protocol. We set the time stamp TIM (i) to indicate the expiration time of i-th CST, after updating the CST and ACT, the counter begins to decrement. If it equals to zero, then the information of the CST will be updated by the explicit sensing mode.

Through using of the NAV, neighbor nodes can recognize the available sub data channel around and the delay time as well. We use the CST (channel status table) to store the status of sub data channels and ACT (available channel table) to record the available ones after the comparison of both source and destination side. 3.3. Channel Status Table (CST) and Available Channel Table (ACT) In order to know the current available sub channel list, each node in the network keeps a channel status table as being showed in TABLE I. There are two kinds of the sensing method to get the status of sub data channel around. One is the explicit sensing; the other one is the implicit sensing. In the explicit sensing mode each node senses its around environment continuously or periodically to update the CST. Whenever a host wants to start the transmission, it has to sense the status of channels utilization including the SINR of the sub data channel that occupied by the incumbent devices as well, store such information into the CST and arrange the corresponding CHRPT consequently. Then, according to the specific pulse responses during the CHRPT period, update the

245

3.4. The Example Procedure of Available Channel List Selection 1. Source node sends RTS to the destination node. The entire procedure is shown in Fig.7. Before the control message transmission, source node has to check the CST to get the available sub-channel list (ASCL) {SC1, SC2, SC3, SC4, and SC5} as that in Fig 7. RTS also include CCNAV for the common control channel reservation and DCNAV for the data channel reservation. Here DCNAV is just a time period that should be delayed not related to any sub data channel. According to Eq.1 and Eq.2, CCNAV can be calculated as Eq.5.
CCNAV = 2 5 PT + 5 SIFS + CHREQ + CTS +

CCNAV D = SIFS + 2 PT + SIFS + CTS

(7)

7. Destination nodes neighbors receive CHREQ and respond with CHRPT-D slots. If certain SCs of CHREQ should be avoided, it means that such set of the SCs has already been occupied by the primary users around the destination nodes neighbors. Neighbor nodes transmit short pulses at the corresponding slot positions to inform the destination node. 8. Destination node decides an optimum sub-channel and transmits CTS. After the responses from the neighbors around the destination, the ACT has been decided. If only j slots left in the ACT which means that only j channels are available currently. Among those j SCs, select one as the sub data channel. In the Fig 8, only one channel (SC5) left in the ACT, destination chooses SC5 and includes it in the CTS, CTS also include DCNAV. After the source node receives the CTS, both side switch to the agreed data channel and begin the packet transmission. Eq. 8 shows the length of the DCNAV included in the CTS.
DCNAV = 2 SIFS + DATA + ACK

(5)

2. Sources neighbor nodes hear RTS and investigate sources ASCL. Because SC1 and SC2 are not available to S-N1 and S-N2 respectively (according to their channel status, incumbent system appeared). Therefore, S-N1 and S-N2 transmit short pulses at 1st and 2nd slots to notify SC1 and SC2 are not available, we use a short pulse to indicate the i-th slot position where the corresponding sub-channel should not be used. 3. Source node resends updated RTS (RTSu). After the response of the SN-1 and SN-2 according to their neighborhood channel status, source node will resend the RTS named RTSu, because of the changing of the CHRPT, CCNAV also changed, and we call it CCNAVu (CCNAV updated). RTSu also includes DCNAV and ASCL {SC3, SC4, and SC5} according to the CST in this case. However, it does not resend RTS if there is no change. According to the Eq. 3.
CCNAVu = 3 PT + 3 SIFS + CHREQ + CTS

(8)

IV. SIMULATION RESULT In this section, we conduct experimental simulations to evaluate the performance of our proposed protocol. We compare the performance of our protocol with that of the DCA [4]. We evaluate the interference time to the incumbent systems. Numbers of the incumbent systems and second users has been arranged in the given area to testify the performance enhancement of our proposed protocol through using the different network configuration. In our simulation environment, there are 10 sub-data channels, one for the transmission of the control packet. All the sub-data channels have the same bandwidth and SINR. Total data transmission time is 3[msec]. DIFS = 50 sec , SIFS =10 sec , backoff time slot = 20 sec , control packet length of RTS and length of CTS: LE N r = LENc = 293 sec (300bits/1Mbps), each channel bandwidth is 1Mbits/sec.We use the ON/OFF model in our simulation which will be described below. Others are similar as that of the DCA protocol

(6)

4. The destination node receives RTS (and RTSu). When the destination node receives the first RTS, it waits until the time period of UIFS is expired. If it could not receive RTSu during UIFS period, the first RTS is handled. Else, the RTSu is handled. 5. Destination node checks sources ASCL and destination nodes CST. According to the CST, SC3 is not available at the destination side. 6. Destination node sends CHREQ to its neighbors. CHREQ includes ASCL that has 2 available sub-channels and CCNAV-D as well. According to Eq. 4.

4.1. ON/OFF Model for Incumbent System Appearance.

246

Because the communication band of the CR is the TV band. Actually, the appearance of signal is as that being showed in the Fig 8. After the appearance of the incumbent devices, channel will be idle for a relatively long time interval. While the type of appearance duration follow the exponential distribution function. We use the Eq. 9 and Eq. 10 to calculate the TimeON (time duration of the incumbent system).
F (t ) = P (TimeON t ) = 1 e t
P= E (TimeON ) E (TimeON ) + E (TimeOFF )

(9)
(10)

the duration of the incumbent device was, the better it will get. In Fig.11, though the 4 lines with different expectation times also increase along with the increment of the probability of incumbents appearance, the interference time to the primary users is reduced greatly compared with the original protocol which is at least 4 times larger than ours with the same network configuration. And due to the characteristics of our proposed mechanism, the longer the duration of TimeON, the larger probability that corresponding channels will be deleted from the CST and ACT consequently. Thus, we can get the better performance.
6
original Proposed

Interference time

Fig.8. ON/OFF model Fig.9 shows the total interference time to the incumbent systems around the neighbors of the communication pair. Here, we use the E (TimeON) = 7, other parameters are the same as that mentioned above. X axis refers to the probability of the appearance of the incumbent devices on sub data channels. And Y axis refers to the interference that impact on the primary users around the neighbor nodes. We compare the proposed protocol with the original protocol DCA, as the probability of the incumbent system appearance increasing. Performance under both protocols are downgraded which is more obviously in the original protocol. The reason is that during the packet transmission period (or more concretely, the channel reservation period), there is no protection mechanism of original protocols to the incumbent users around the communication pair; thus, interference to the incumbent devices around of communication pair can not be controlled. While in our proposed protocol, the communication pairs already know the sub channel sets that have been occupied by other users through CHRPT period reported by its neighbors. Interference is not zero is because of the appearance of incumbent users during the data transmission time which we can not avoid. Apparently, our new protocol can reduce the impact on the primary users around neighborhood greatly as that being showed in the Fig.9. In order to test the performance under the different circumstance, we present performance of our proposed protocol with four different expectation times TimeON as well. Fig.10 shows the total interference time with original protocol that does not have the mechanism to protect the incumbent users around the neighborhood of communication pair. As we can see in the graph, along with the increasing of the probability of the incumbent devices appearance, the longer

0 0.1 0.2 0.3 Probability 0.4 0.5 0.6

Fig. 9. Total interference time to the incumbent devices around the neighbor of communication pair
5.5 5.3 5.1 Interference time 4.9 4.7 4.5 4.3 4.1 3.9 3.7 3.5 0.1 0.2 0.3 Probability 0.4 0.5 0.6

E(TimeON)=7msec E(TimeON)=5msec E(TimeON)=3msec E(TimeON)=1msec

Fig. 10. Interference time of the original protocol We use the different simulation parameters to testify the superiority of our proposed protocol. Fig. 12, 13 and 14 illustrates the performance using the different number of neighbor nodes. The proposed approach performs quite good compare with the original one. We set the total transmission time to 3msec, E (TimeON) = 7. It is clear that as the number of the neighbor nodes increases from 1 to 3. Though the interference on the primary node is getting greater as well, compared to the original mechanism only 30 percentage of the impact we have on the incumbent users around. Because the increment number of neighbor node will also increase the total sub data channel amount, though we can eliminate some of

247

interference time

them by using the CHRPT mechanism in our protocol, the interference that will impact on the incumbent devices during the data transmission will increase as well. As we can see from the figures described, our proposed protocol can get a better performance compared with the original DCA protocol in the aspect of interference to the incumbent devices.
1.2

0.6

Original protocol
0.5

Proposed protocol

0.4

0.3

0.2

E(TimeON)=7msec E(TimeON)=5msec E(TimeON)=3msec E(TimeON)=1msec

0.1

Interference time

0.8

0 0.1 0.2 0.3 Probability 0.4 0.5 0.6

0.6

Fig. 14. Number of neighbor nodes = 3


0.4 0.2

0 0.1 0.2 0.3 Probability 0.4 0.5 0.6

V. CONCLUSION AND FUTRURE WORK In this paper, we have addressed the hidden incumbent device problem in the multi-channel environment with CR technology, and we proposed a distributed coordinated spectrum sharing MAC protocol for cognitive radio wireless ad hoc networks, which operating over the open spectrum to avoid the hidden incumbent node problem mainly through collecting the primary users information of neighbor node to the communication pair. During the CHRPT period, each node near the communication pair can report their channel status to the pair that about to begin the packet transmission by using the pulse to avoid the broadcast storm. Simulation result shows that the new protocol can solve the hidden incumbent node problem successfully while only a little degradation on the access delay. However, in this protocol, of primary concern in this study are single hop issues regarding the avoidance of interference to incumbent devices near the source and destination node. A method to extend this mechanism to the multi-hop routing path selection remains as a future work worth considering. In addition, only the SINR is considered in the channel selection among the remaining available channels; for optimal channel utilization, policy channel selection mechanisms based on alternative channel situations and topology are also considered necessary in future works. VI. REFERENCE
[1] [2] [3] FCC, Spectrum policy task force report, No. 02-155,Novermber 2002 Simon Haykin, Life Fellow, IEEE, Cognitive Radio: Brain-Empowered Wireless Communications IEEE 802.22-05/0007r47,Functional Requirements for the 802.22 WRAN standard,2006 S.-L. Wu, C.-Y. Lin, Y.-C. Tseng and J.-P. Sheu, A New Multi-Channel MAC Protocol with On-Demand Channel Assignment for Multi-Hop Mobile Ad Hoc Networks, in Intl Symposium on Parallel Architectures, Algorithms and Networks (I-SPAN) , 2000. Jungmin So, Nitin Vaidya, multi-Channel MAC for Ad Hoc Networks: Handling Multi-Channel Hidden Terminals Using A Single Transceiver

Fig. 11. Interference time of the proposed protocol


0.18 0.16 0.14
original protocol

Proposed protocol

Interference time

0.12 0.1 0.08 0.06 0.04 0.02 0 0.1 0.2 0.3 Probability 0.4 0.5 0.6

Fig. 12. The interference time to the incumbent devices around the neighbor of communication pair with the average number of neighbor nodes = 1
0.35 0.3 0.25 Interference time 0.2 0.15 0.1 0.05 0 0.1 0.2 0.3 Probability 0.4 0.5 0.6

Original protocol Proposed protocol

[4]

Fig. 13. Number of neighbor nodes = 2

[5]

248

[6]

J. Deng and Z. Haas, Dual Busy Tone Multiple Access (DBTMA): A New Medium Access Control for Packet Radio Networks, in Proc. Of IEEE ICUPC, Florence, Italy, 1998. [7] A. Nasipuri, J. Zhuang and S. R. Das, A Multichannel CSMA MAC Protocol for Multihop Wireless Networks, in Proc. of IEEE Wireless Communications and Networking Conference (WCNC), September 1999.

249

Das könnte Ihnen auch gefallen