Efficient Resource Allocation for Rapid Link Recovery ... - IEEE Xplore

524

IEEE Transactions on Consumer Electronics, Vol. 56, No. 2, May 2010

Efficient Resource Allocation for Rapid Link Recovery and Visibility in Visible-Light Local Area Networks Woo-Chan Kim, Chi-Sung Bae, Soo-Yong Jeon, Sung-Yeop Pyun, Student Member, IEEE, and Dong-Ho Cho, Senior Member, IEEE Abstract — Visible-light communication provides many

advantages over other forms of communications, such as visibility, high SNR (Signal to Noise Ratio), easy installation, freedom of interference from radio or electromagnetic waves, usage of license free frequency band, and high security. Furthermore, the exponentially increasing requirements and quality of LEDs (Light Emitting Diodes) have encouraged the development of visible-light communication that makes use of LEDs. In a visible-light communication system, an LOS (Line of Sight) link between two transceivers should be guaranteed due to the straightness of the visible-light signal. However, link failure caused by temporary blocking or poor orientation of a transmitter frequently occurs, causing burst frame errors. In this paper, we focus on how to let a user know about link failure as quickly as possible, in order that the link failure problem may be solved by realigning the transmission signal towards the receiver. This is furthermore a relatively straightforward matter, by the very nature of this system being a visible means of communication. It is also required that the visibility should be supported when a user attempts initial access procedure. Nevertheless, by supporting visibility, system performance can be degraded because some additional resources are required for the visibility to be enabled. In this paper, we propose three schemes for supporting visibility without reducing system performance in a visible-light local area network1. Index Terms — VLC, VLAN, optical wireless communication, LED, link recovery, visibility.

I. INTRODUCTION LEDs (Light Emitting Diodes) are considered by many researchers to represent the next generation of lighting technology. The lamp of an LED has some excellent features, such as low power consumption, long life time, small size, high density, rapid response time, and low cost. It can also be used in many applications, such as in color displays, traffic lights, notice boards, automobiles, and cellular phones. Applications using white LEDs have in particular been the subject of much attention of late [1], [2]. Japan and other advanced countries have suggested that white LED technology may be considered to be the next generation of lighting technology. Furthermore, international agreements such as the 1

The authors are with the department of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Korea (email: [email protected], [email protected], [email protected], [email protected], [email protected]). Contributed Paper Manuscript received November 12, 2009 Current version published 06 29 2010; Electronic version published 07 06 2010.

Kyoto Protocol, WEEE (Waste Electrical and Electronic Equipment Directive) and RoHS (Restriction of Hazardous Substances Directive) have encouraged the use of LEDs. With such efforts, existing fluorescent or incandescent lamps could be replaced with LED lamps within a few years. The promotion of LEDs has led to the emergence and development of VLC (Visible-Light Communication). In VLC, LEDs are used both as communication and lighting devices. Hence, the VLC system is a kind of optical wireless communication system that uses visible-light as a transmission medium. VLC enjoys many advantages compared with the current wireless communication systems of IR (Infrared) and radio wave wireless communication. IR communication is one of the most widely used communication systems in PANs (Personal Area Networks), but its use is hazardous to the human eye, and can be characterized by a dangerously high energy density, as a result of its invisibility. A high data transmission rate cannot therefore be achieved using IR communication. In contrast, VLC is safe for the human eye because it is visible, and provides a high rate of data transmission. Furthermore, a high SNR (Signal to Noise Ratio) may be achieved using a power of only a few Watts, due to the high energy efficiency of the LED. The other alternative of a wireless communication system using radio waves, such as a cellular system or wireless LAN, enjoys widespread use. However, this technology commonly suffers from a frequency allocation problem due to a lack of available radio frequencies, and the electromagnetic waves used can cause biological damage to humans. Compared with radio wave wireless communication devices, the use of VLC is harmless to humans and does not require a license. Furthermore, it does not cause the malfunction of aircraft equipment or medical instruments because visible-light signals do not interfere with radio waves. Besides these advantages, VLC provides high security due to its visibility, and is easy to install if LED lamps are already being used. The use of VLC has been proposed and mainly researched by Nakagawa laboratory at Keio University. They have evaluated some basic performance parameters using numerical analysis and simulations [3]-[7], and have researched the integration of the system with PLC (Power Line Communication) [8] and the OFDM (Orthogonal Frequency Division Multiplexing) module [9], [10], and have investigated the effects of interference, reflection, and shadowing [11], [12]. They have also investigated the modulation method for VLC [13], [14], and proposed applications using visible-light such as a positioning and

0098 3063/10/$20.00 © 2010 IEEE

W.-C. Kim et al.: Efficient Resource Allocation for Rapid Link Recovery and Visibility in Visible-Light Local Area Networks

tracking system [15], [16]. Furthermore, the VLCC (Visible Light Communications Consortium) was established with the objective of generating publicity for, and achieving the standardization of, VLC using LEDs in Japan. The standardization of VLC is also in progress. The IEEE 802.15 VLC was chartered as an IG (Interest Group) in November 2007, and promoted to an SG (Study Group) in March 2008. It was then promoted further to a TG (Task Group) in January 2009, and the IEEE standard for VLC PHY and MAC will be completed by 2011. Existing research and standards for the VLC are focused on the physical layer. However, a MAC (Medium Access Control) protocol is required for the VLC, because the physical characteristics of visible-light are different from those of radio waves or IR [17], [18]. We herein consider a new MAC protocol for VLC that considers the physical characteristics of visible-light. In the MAC protocol, we define three service modes, namely the VL (Visible-light Local Area Network) mode, the BI (Broadcast Information) mode and the PI (Peripheral Interface) mode. Fig. 1 shows examples of each service mode. In the VL mode, several MNs communicate with an AP (Access Point). An AP broadcasts data to MNs in the IB mode, and MNs communicate directly through a P2P (Peer To Peer) link in the PI mode.

525

recovery and visibility. Section V evaluates the performance of the proposed schemes. Finally, we present some conclusions to the paper in Section VI. II. MAC PROTOCOL FOR THE VLAN SYSTEM The VLAN system is synchronous and time-slotted. Fig. 2 shows a frame structure for the VLAN system. A frame is divided into several timeslots. The durations of the frame and the slot are fixed, and denoted as Tf and Ts, respectively. The VLAN system supports both full and half duplex modes. In the case of the full duplex mode, a timeslot is simultaneously allocated for the transmission of uplink and downlink data, because the data transmissions of the uplink and downlink do not interfere with each other, as a result of the straightness of visible-light. In the case of the half duplex mode, a timeslot is allocated for either uplink or downlink transmission. Therefore, the throughput of the full duplex mode is double that of the half duplex mode.

Fig. 2. Frame structure for the VLAN system (a) full duplex mode (b) half duplex mode Fig. 1. Possible service modes for a VLC system

When operated in the VL mode, the VLC system is known as a VLAN (Visible-light Local Area Network). In a VLAN system, users must point a transmission signal towards the receiver of an AP correctly, because an MN transmits sharp visible-light signals for high energy efficiency. Furthermore, the communication link may be terminated relatively easily by temporary blocking (e.g. by people or paper), due to the straightness of the visible-light signal. In order to solve this problem, we propose a rapid link recovery scheme, which enables a user to recognize the disconnection of a link as soon as possible, so that the communication link may be reestablished. If the user can see the signal, re-establishing the link is a relatively straightforward matter. In order to provide visibility, extra resources should be allocated to MNs, which degrades their performance. Furthermore, visibility cannot be supported if the uplink resources are insufficient. We therefore propose a resource allocation scheme for visibility in order to overcome these problems. The remainder of this paper is organized as follows. In Section II, a new MAC protocol for the VLAN system is briefly described. Section III and Section IV present and analyze proposed resource allocation schemes for rapid link

The first downlink slot is used for the transmission of the frame start-flag (FS) and the frame header (FH). The FS indicates the start of each frame and is used for the synchronization of the MNs. The FH includes the identification number, the service type and capability of an AP, and all the control information on slot usage and allocation. In receiving the FH, each MN obtains all the information needed for the transmission and receipt of data, such as the service type (VLAN, peer-to-peer, broadcast), the usage of the timeslot (broadcast, multicast, unicast), the schedule of forthcoming data transmission and receipt, and the destination. The FS and the FH are transmitted using the lowest data transmission rate, because their transmission must be as robust as possible. The first uplink slot is used as a contention slot (C-slot). The C-slot consists of several mini-slots whose duration is fixed at Tm. An MN that attempts initial access to an AP randomly selects a mini-slot and transmits signals at the corresponding mini-slot. Some parts of the mini-slots can be allocated for fast feedback of MNs in a dedicated fashion. Since usage information for each mini-slot is included in the FH, the MNs can make proper use of the mini-slot without confusion.

526


III. PROPOSED RESOURCE ALLOCATION SCHEME In order to solve the problem of the frequent disconnection of the link, an AP rapidly reports any link disconnection to users, in order that users can re-establish their communication link. To achieve this, the AP should allocate sufficient resources for managing the status of the link connection of each MN. However, the system performance may deteriorate as a result of the extra resources allocated for rapid link recovery. Besides this, additional resources are allocated to users to support visibility, which also affects the performance of the system. We therefore propose resource allocation schemes for rapid link recovery and visibility that minimize performance degradation. A. Rapid Link Recovery Scheme In a VLAN system, the LOS (Line of Sight) link must be guaranteed because visible-light cannot pass through solid obstacles like walls. However, temporary blockages (e.g. people waking between the transmitter and receiver) cause burst frame errors, which occur frequently in VLAN systems. In addition, the poor orientation of an MN causes a decrease in the SNR or even link disconnection. In order to solve these problems, an AP provides a rapid link recovery service in the case of link failure. Fig. 3 shows the flow chart for the rapid link recovery scheme using a dedicated mini-slot. If an AP detects an initial access signal, it allocates a dedicated minislot to the corresponding MN. Then, the MN transmits signals at the dedicated mini-slot until it becomes disassociated with the AP. If the AP receives the first signal at the dedicated mini-slot, it recognizes uplink synchronization and allocates resources for the association. If the AP does not receive signals at the dedicated slot for a fixed number of consecutive frames, it determines that the VLC link has become disconnected, and transmits an UL-UNSYNC message to the MN. When the MN receives this message, it stops transmitting, apart from the status signal transmission at the dedicated minislot, until the AP receives the signal transmitted by the MN. In case of disconnection as a result of poor orientation, users must set up the direction of uplink signal to the AP receiver correctly. If the disconnected VLC link is recovered, and the AP can recognize uplink synchronization by the signal at the dedicated mini-slot, the AP transmits an UL-SYNC message to the MN. After the MN receives this message, it can then transmit data to the AP. B. Visibility Supporting Scheme The VLAN system is a synchronous time-slotted system, thus only one timeslot might be assigned to an MN that has a low priority. Furthermore, an AP cannot allocate very many timeslots to each MN if there are too many MNs. The MN transmits visible-light signals only at the allocated timeslot. Hence, users cannot see the uplink signal if they use too few timeslots, because the time for which the light is off is much longer than that for which the light is on in this case. However, visibility is positively necessary in the following three cases. The first case occurs when an MN attempts an initial access. In this case, it is very hard for a user to point the uplink signal

towards the AP receiver correctly, because the MN has never connected to the AP. A user can set up the direction of the uplink signal easily if he/she can see the uplink signal. The second case occurs when the VLC link between an AP and an MN has become disconnected due to poor orientation. In order to re-establish the VLC link, the user should re-orient the uplink signal towards the AP receiver. In this case, visibility helps the user to point the uplink signal towards the AP receiver correctly. The final case occurs when the VLC link between an AP and an MN becomes disconnected due to obstacles. In this case, the user might not be aware of the reason why the VLC link is disconnected. If visibility is supported, users can identify the nature of the obstacles that is interrupting the communication in the VLC link, and solve the problem by removing the obstacle or moving to another location in which the LOS is guaranteed.

Fig. 3. Flow chart for rapid link recovery scheme

In order to support visibility in the three cases described above, an MN should transmit extra signals other than the data signal. These extra signals for visibility are herein termed “point-shot signals”. A point-shot signal consists of a signal “1” for visibility and contains no other information. We now consider three schemes according to a resource allocation method for point-shot signals. Firstly, an MN can transmit a point-shot signal using specific timeslots. The specific timeslot is used for both the data signal and the point-shot signal. Fig. 4 shows an example of a specific point-shot scheme. This scheme does not reduce channel utilization because whole timeslots can be used for the data slot if there is no point-shot signal. However, there will be a collision in timeslot 4, which is used by MN 3 if MN 4 transmits the point-shot signal to timeslot 4 in Fig. 4. In this case, MN 3 cannot transmit data to the AP due to collision with the point-shot signal of MN 4. Thus, this scheme reduces the success rate of the transmission.


527

Fig. 4. An example of a specific slot for a point-shot signal

Secondly, an MN can transmit a point-shot signal using reserved slots that are used only for the point-shot signal. Fig. 5 shows an example of a reserved point-shot scheme. This scheme causes no collision to occur, since the slots for the data signal and the point-shot signal are divided separately. A high transmission success rate is therefore guaranteed under this scheme. However, it reduces the channel utilization, because the reserved slot cannot be used even when no point-shot signal is present.

Fig. 5. An example of a reserved slot for a point-shot signal

Finally, an MN can transmit the point-shot signal using semi-reserved slots. The semi-reserved slot is used for both the data signal and the point-shot signal, but it is different from the specific slot. In contrast to the specific slot, the semireserved slot is used for the data signal only if there is no MN transmitting a point-shot signal. Fig. 6 shows an example of a semi-reserved point-shot scheme. In Fig. 6, there will be a collision in slot 4 that is used by MN 3 if MN 4 transmits the point-shot signal to slot 4. MN 3 then recognizes the consecutive frame errors and stops transmitting data to avoid the collision. Thus, this scheme seldom decreases the success rate of the transmission. However, an MN that uses the semi-reserved slot cannot transmit a data signal as long as any other MNs are transmitting a point-shot signal to the semi-reserved slot. Therefore, a semi-reserved slot is allocated to an MN that has low priority. The semi-reserved slot may also be allocated in cases where uplink resources are empty or where the AP allocates uplink resources using a polling method.

Fig. 6. An example of a semi-reserved slot for a point-shot signal

IV. NUMERICAL ANALYSIS A. Model Assumptions We assume that there are n timeslots and that the first timeslot consists of m mini-slots. Among the n timeslots, l timeslots are used for the point-shot signal. We also assume that arrival process is a bulk Poisson process with intensity λ and fixed bulk size N. The N MNs that arrive simultaneously attempt an initial access, and a specific MN connects to an AP. The AP allocates one timeslot to the MN. This is valid for a VoIP or a streaming service where a minimum service rate is significant. MNs are served according to an exponential service time distribution with parameter μ. B. Rapid Link Recovery Scheme We use a Markov Chain model to analyze a proposed scheme. Let the state of the Markov Chain be the number of MNs in the VLAN system. For simplicity, we neglect the point-shot signal in this subsection, i.e. l = 0. We may then model the Markov Chain as shown in Fig. 7.

Fig. 7. State diagram of Markov Chain model

In Fig. 7, Ps(k) is the probability that a specific MN among the N incoming MNs succeeds in an initial access where there are k MNs in the VLAN system. To derive Ps(k), we define yk as the number of mini-slots that can be used for the initial access where k uplink slots are allocated to MNs. If the VLAN system does not provide a rapid link recovery service, yk is m regardless of the value of k. However, yk is inversely proportional to k since one mini-slot is assigned to each MN that connects to an AP in the rapid link recovery scheme.

528


Therefore, yk is (m-k) if an AP provides the rapid link recovery service. We can then derive Ps(k) as follows. (k ) s

P

⎛ y −1⎞ ⎟⎟ = ⎜⎜ k ⎝ yk ⎠

N −1

(1)

n −1

In Fig. 7, πi, the probability that the number of MNs in the VLAN system is i is as follows, using the balance equation. i

1 ⎛ λ ⎞ i −1 π i = ⎜⎜ ⎟⎟ ∏ Ps( j ) ⋅ π 0 i ! ⎝ μ ⎠ j =0

⎡ n −1 1 ⎛ λ ⎞i i −1 ( j ) ⎤ π 0 = ⎢1 + ∑ ⎜⎜ ⎟⎟ ∏ Ps ⎥ ⎢⎣ i =1 i ! ⎝ μ ⎠ j = 0 ⎥⎦

(2)

n −1

{

−1

(3)

}

(4)

i =0

Let Pdrop be the probability that an MN cannot connect to an AP due to the collision of the initial access or the full use of resources. Then, Pdrop is, n −1

{

}

{

}

γ ' = ∑ 1 − (1 − τ ⋅ γ )i ⋅ π i i =0

(7)

Furthermore, the initial access rate of all the MNs in the system, λ’, is described as follows.

λ ' = ( 1 − Pdrop ) f ⋅ λ

Here, the VLAN system operates even when the traffic intensity ρ = λ/(n-1)μ is greater than 1, because an MN leaves the VLAN system if there is no empty slot. We show the collision probability of initial access, the drop probability, and the expected channel utilization, which are expressed as Pcol, Pdrop, and Uch, respectively. Then, Pcol can be calculated using (1) as follows.

Pcol = ∑ π i ⋅ 1 − Ps( i )

communication or if the user does not carry out the orientation properly. If the reorientation rate of an MN is γ and it takes τ seconds to recover the link failure, the expected reorientation rate of all the MNs in the system, γ’, is given by,

Pdrop = ∑ π i ⋅ 1 − Ps( i ) + π n −1 ⋅ Ps( n −1)

(5)

(8)

where f is the frame length. In a specific point-shot scheme, data transmission can be interrupted by a point-shot signal whenever one is transmitted. The uplink slot utilization rate, Uspc, and data transmission success rate, Sspc, in this scheme may therefore be expressed as, n −1

U spc = { 1 − (λ '+γ ')}⋅ ∑ i ⋅ π i

(n − 1)

i =0

⎧n −1− l ⎫ ⎪ ∑i ⋅πi + ⎪ ⎪ i =0 ⎪ + (λ '+γ ') ⋅ ⎨ n −1 ⎬ ⎪ (n − 1 − l ) ⋅ π ⎪ i ⎪⎩i ∑ ⎪⎭ = n −l

(n − 1)

⎡ n −1 ⎤ ⎢∑ i ⋅ π i − (λ '+γ ') ⋅ ⎥ i =0 ⎢ ⎥ = n −1 ⎢ ⎥ {i − (n − 1 − l ) ⋅ π i }⎥ ∑ ⎢ i = n −l ⎣ ⎦

(n − 1)

(9)

i =0

Also, channel utilization, Uch, may be calculated using (2) and (3) as follows. n −1

U ch = ∑ i ⋅ π i

(n − 1)

(6)

i =0

C. Visibility Supporting Scheme We have presented three schemes for visibility using point-shot signals. In this section, we will compare the three schemes in respect of channel utilization and transmission success rate. As previously mentioned, it is not necessary for MNs to transmit the point-shot signal all the time. Instead, the MNs transmit the point-shot signal in the case of an initial access or for reorientation. Reorientation is required when something interrupts

S spc = {1 − (λ '+γ ')}⋅ 1 n −1 ⎛ n −1− l ⎞ n −1− l + (λ '+γ ') ⋅ ⎜ ∑ π i + ∑ ⋅πi ⎟ i i = n −l ⎝ i =0 ⎠ = 1 − (λ '+γ ') ⋅

(10)

n −1 ⎧ ⎛ n −1− l ⎞⎫ n −1− l ⋅ π i ⎟⎬ ⎨1 − ⎜ ∑ π i + ∑ i i = n −l ⎠⎭ ⎩ ⎝ i =0

= 1 − (λ '+γ ') ⋅

i − n +1+ l ⋅πi i i = n −l n −1

∑

In a reserved point-shot scheme, the data and point-shot slots are separated. Then, the uplink slot utilization rate, Ursv, and the data transmission success rate, Srsv, in this scheme may


be derived as,

0.3

S rsv = 1

(11)

(12)

In a semi-reserved point-shot scheme, we assume that an MN that is using the semi-allocated slot stops transmitting data when it recognizes d consecutive errors. Let Usm and Ssm respectively be the uplink utilization and data transmission success rates in this scheme. Usm is the same as Uspc since l slots cannot be used while the point-shot signal is being transmitted in either scheme. In a semi-reserved point-shot scheme, the transmission failure rate from the reorientation of all the MNs in the system, γ”, may be described as, n −1

{

}

γ ' ' = ∑ 1 − (1 − df ⋅ γ )i ⋅ π i

0.25

0.2

0.15

0.1

0.05 0.2

0.8

1

1.2

1.4

Fig. 8. Collision probability vs. traffic intensity

(13)

0.7

Drop probability, P

(14)

S sm = {1 − (λ '+γ ' ')}⋅ 1

0.6 0.5

conv (N=3) prop (N=3) conv (N=4) prop (N=4) conv (N=5) prop (N=5)

0.4 0.3 0.2 0.1 0.2

(15)

V. NUMERICAL & SIMULATION RESULTS In this section, we show results from the performance evaluation of a proposed rapid link recovery and visibility supporting scheme. We assume that there are 10 uplink slots and 20 mini-slots. Also, 3 to 5 MNs arrive simultaneously at a frequency ranging from every 10 seconds to every 2 minutes, and the frame length is 10 ms. The average communication time is 5 minutes. Firstly, we show the performance of the rapid link recovery scheme. This scheme ensures that an AP rapidly recognizes link failure. However, the collision probability of an initial access increases, because the mini-slots are used for the rapid link recovery instead of for the initial access. Fig. 8 shows the collision probability of the initial access in relation to traffic intensity ρ. The rapid link recovery scheme increases the collision probability of the initial access by up to 35% for ρ = 1. Furthermore, the collision probability of the initial access

0.6

0.8

1

1.2

1.4

Fig. 9. Drop probability vs. traffic intensity

1 0.9 ch

n −1 ⎧ ⎛ n −1− l ⎞⎫ n −1− l 1 − π + ⋅ π i ⎟⎬ ⎨ ⎜∑ i ∑ i i = n −l ⎠⎭ ⎩ ⎝ i =0 n −1 i − n +1+ l = 1 − (λ '+γ ' ') ⋅ ∑ ⋅πi i i = n −l

0.4

Traffic intensity, ρ

Channel utilization, U

n −1 ⎛ n −1− l ⎞ n −1− l ⋅πi ⎟ + (λ '+γ ' ') ⋅ ⎜ ∑ π i + ∑ i i = n −l ⎝ i =0 ⎠ = 1 − (λ '+γ ' ') ⋅

0.6

0.8

Hence, Usm and Ssm are derived as follows.

U sm = U spc

0.4


drop

i =0

conv (N=3) prop (N=3) conv (N=4) prop (N=4) conv (N=5) prop (N=5)

col

n −1 ⎫ ⎧n −1− l = ⎨ ∑ i ⋅ π i + ∑ (n − 1 − l ) ⋅ π i ⎬ (n − 1) i = n −l ⎭ ⎩ i =0

Collision probability, P

U rsv

529

conv (N=4) prop (N=4)

0.8 0.7 0.6 0.5 0.4

0.2

0.4

0.6

0.8

1

1.2

1.4


Fig. 10. Channel utilization vs. traffic intensity

increases as the traffic intensity increases. However, for intense traffic conditions, an MN can fail in the initial access due to a lack of uplink resources as well as collision of the initial access. The drop probability (i.e. the probability that an MN fails during the initial access) is therefore a more important performance indicator than the collision probability.

530


collisions between the data and point-shot signals in the specific point-shot scheme. Fig. 12 shows the channel utilization with respect to traffic intensity ρ. In this case, the performance of the reserved pointshot scheme is worse than that of the other two schemes. Extra timeslots for the point-shot signal are required, which reduces channel utilization. This represents a serious problem in the reserved point-shot scheme because the timeslots for the point-shot signal cannot be used for the data signal.

1 Specific slot Reserved slot Semi-reserved slot

Transmission success rate

0.99 0.98 0.97 0.96 0.95 0.94 0.93

IV. CONCLUSION

0.92

In this paper, we have proposed a rapid link recovery scheme and three visibility supporting schemes for a VLAN system. The proposed rapid link recovery scheme helps the rapid recovery of the link failure by making an AP manage the connection status. However, it increases the collision probability of an initial access, because the number of minislots that can be used for the initial access decreases. However, the increased collision probability does not reduce channel utilization or system throughput as shown in the results. Furthermore, we have presented three schemes to support visibility. The data transmission success rate is low in a specific point-shot scheme and channel utilization is poor in a reserved point-shot scheme. The results in a semi-reserved point-shot scheme show good performance in these two metrics, however. We conclude that the semi-reserved pointshot scheme represents the best approach. In the future, interference from sunlight and other LED lights should be avoided, and the effect of reflection and shadowing should also be considered. We also think that MIMO (Multiple Input Multiple Output) transmission could represent a solution to increase the throughput of these systems.

0.91 0

0.2

0.4

0.6

0.8

1

1.2

1.4


Fig. 11. Transmission success rate vs. traffic intensity 0.9 0.8

Specific slot Reserved slot Semi-reserved slot

Channel utilization

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

0.2

0.4

0.6

0.8

1

1.2

1.4


Fig. 12. Channel utilization vs. traffic intensity

Fig. 9 shows the drop probability in relation to the traffic intensity ρ. The rapid link recovery scheme increases the drop probability by up to 10% when ρ = 1. However, the drop probability is affected less by the rapid link recovery scheme compared to the collision probability of the initial access. This is because the initial access failure is influenced by the lack of uplink resources greater than the collision of the initial access as the traffic is more intense. Fig. 10 shows the channel utilization in relation to the traffic intensity ρ. The rapid link recovery scheme does not decrease channel utilization. In fact, channel utilization is much more important than collision probability or drop probability, because it directly affects the system performance such as throughput. The rapid link recovery scheme will therefore not decrease the system throughput. We now describe the performance of the visibility supporting schemes. We have fixed the reorientation rate γ at 1/60, and the link recovery time τ at 3 seconds. Also, an MN stops data transmission if it recognizes five consecutive errors in the semi-reserved point-shot scheme. Fig. 11 shows the data transmission success rate with respect to traffic intensity ρ. The performance of the specific point-shot scheme is inferior to that of the other two schemes. The performance degradation increases with traffic intensity, because high traffic causes

ACKNOWLEDGMENT This work was supported by the IT R&D program of MKE/KEIT [2008-F-004-02, 5G mobile communication systems based on beam division multiple access and relays with group cooperation]. REFERENCES [1] [2] [3] [4] [5] [6] [7]

Y. Gu, N. Narendran, and J. P. Freyssinier, “White LED performance,” in Proc. of SPIE, vol. 5530, pp. 119-124, Oct. 2004. A. Okuno, Y. Miyawake, N. Oyama, and W. Dongxu, “Unique white LED packaging systems,” in Proc. IEEE EMAP, pp. 1-5, Dec. 2006. Y. Tanaka, S. Haruyama, and M. Nakagawa, “Wireless optical transmission with white colored LED for wireless home links,” in Proc. IEEE PIMRC, vol. 2, pp. 1325-1329, Sept. 2000. Y. Tanaka, T. Komine, S. Haruyama, and M. Nakagawa, “Indoor visible communication utilizing plural white LEDs as lighting,” in Proc. IEEE PIMRC, vol. 2, pp. F81-F85, Sept. 2001. T. Komine, Y. Tanaka, S. Haruyama, and M. Nakagawa, “Basic study on visible-light communication using light emitting diode illumination,” in Proc. ISMOT, pp. 45-48, June 2001. Y. Tanaka, T. Komine, S. Haruyama, and M. Nakagawa, “Indoor visible light data transmission system utilizing white LED lights,” IEICE Trans. on Communications, vol. E86-B, no. 8, pp. 2440-2454, Aug. 2003. T. Komine and M. Nakagawa, “Performance evaluation of visible-light wireless communication system using white LED lightings,” in Proc. IEEE ISCC, vol. 1, pp. 258-263, June 2004.

W.-C. Kim et al.: Efficient Resource Allocation for Rapid Link Recovery and Visibility in Visible-Light Local Area Networks [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18]

T. Komine and M. Nakagawa, “Integrated system of white LED visible-light communication and power-line communication,” IEEE Trans. on Consumer Electronics, vol. 49, no. 1, pp. 71-79, Feb. 2003. Y. Tanaka, T. Komine, S. Haruyama, and M. Nakagawa, “A basic study of optical OFDM system for indoor visible communication utilizing plural white LEDs as lighting,” in Proc. ISMOT, pp. 303-306, June 2001. T. Komine, S. Haruyama, and M. Nakagawa, “Performance evaluation of narrowband OFDM on integrated system of power line communication and visible light wireless communication,” in Proc. ISWPC, pp. 6-11, Jan. 2006. T. Komine and M. Nakagawa, “Fundamental analysis for visible-light communication system using LED lights,” IEEE Trans. on Consumer Electronics, vol. 50, no. 1, pp. 100-107, Feb. 2004. T. Komine and M. Nakagawa, “A study of shadowing on indoor visible-light wireless communication utilizing plural white LED lightings,” in Proc. ISWCS, pp. 36-40, Sept. 2004. H. Sugiyama, S. Haruyama, and M. Nakagawa, “Experimental investigation of modulation method for visible-light communications,” IEICE Trans. Communications, vol. E89-B, no. 12, pp. 3393-3400, Dec. 2006. H. Sugiyama, S. Haruyama, and M. Nakagawa, “Brightness control methods for illumination and visible-light communication systems,” in Proc. IEEE Wireless and Mobile Communications, pp. 78-83, March 2007. M. Yoshino, S. Haruyama, and M. Nakagawa, “High-accuracy positioning system using visible LED lights and image sensor,” in Proc. IEEE Radio and Wireless Symposium, pp. 439-442, Jan. 2008. T. Saito, S. Haruyama, and M. Nakagawa, “A new tracking method using image sensor and photo diode for visible light road-to-vehicle communication,” in Proc. IEEE ACT, pp. 673-678, Feb. 2008. D. H. Cho, B. C. Jung, C. S. Bae, and W. C. Kim, “MAC requirements for visible light communication systems,” IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs) Std., Sept. 2008. T. G. Kang, S. K. Lim, D. H. Kim, K. H. Lee, T. W. Kim, M. A. Chung, and S. W. Sohn, “VLC MAC considerations,” IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs) Std., Sept. 2008.

BIOGRAPHIES Woo-Chan Kim received a B.S. degree in Electrical Engineering from Korea University, Seoul, Korea, in 2007. He is currently pursuing a Ph.D. degree in Electrical Engineering from the Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Korea. His research interests include wireless visible-light communication systems and next-generation cellular systems.

531

Chi-Sung Bae received a B.S. degree in Electrical Engineering from the Pohang University of Science and Technology, Pohang, Korea, in 2000. He received an M.S. degree in Electrical Engineering from the Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Korea, in 2004. He is currently pursuing a Ph.D. degree in KAIST. His research interests include resource allocation and multi-hop transmission techniques for next-generation wireless systems. Soo-Yong Jeon received B.S., M.S., and Ph.D. degrees in Electrical Engineering from the Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Korea, in 2003, 2005, and 2009, respectively. He is currently a postdoctoral fellow in Electrical Engineering in KAIST. His research interests include wireless visible-light communication systems and wireless multi-hop communication systems. Sung-Yeop Pyun received a B.S. degree in Information and Communication Engineering from Sunmoon University, ChungNam, Korea, in 2006. He received an M.S. degree in Electrical Engineering from the Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Korea, in 2008. He is currently pursuing a Ph.D. degree in Electrical Engineering in KAIST. His research interests include wireless visiblelight communication systems and next-generation cellular systems. Dong-Ho Cho (SM’85) received a B.S. degree in Electrical Engineering from Seoul National University, in 1979 and M.S. and Ph.D. degrees in Electrical and Electronics engineering from the Korea Advanced Institute of Science and Technology (KAIST), in 1981 and 1985, respectively. Between 1987 and 1997, he was a Professor of Computer Engineering at Kyunghee University. Since 1998, he has been a Professor of Electrical Engineering at KAIST. He has been working as a director of the KAIST Institute for IT convergence since 2007. His research interests include wired/wireless communication networks, protocol, and services. He has published 120 international journal papers and 195 international conference papers.