Content uploaded by Gang Wang
Author content
All content in this area was uploaded by Gang Wang on Dec 06, 2019
Content may be subject to copyright.
A Novel MAC Protocol for Wireless Network using
Multi-Beam Directional Antennas
Gang Wang, Peng Xiao, Wenming Li
Dept. of Computer Science and Engineering, University of Connecticut, CT, USA
Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China
Email: gang.wang@uconn.edu g.wang.china86@gmail.com
Abstract—The use of directional antennas in wireless networks
has received growing attention because of its high spatial reuse
and high antenna gains. The Medium Access Control(MAC)
protocol with directional antennas is nontrivial due to the limita-
tions in wireless environment. The existing protocols commonly
assume that the nodes can operate in both directional and
omni-directional modes. In this paper, we propose a new MAC
protocol for multi-beam directional antennas. In the protocol,
each beam-sector has its own control channel, thus the commu-
nications among different beam-sectors are independent. It uses
the directional network allocation vector(DNAV) to record the
establishment processes. After sensing all the sectors of the multi-
beam antenna, it uses the global assignment strategy to assign the
directional communication channels. Extensive simulation results
show that our protocol can significantly improve the throughput
of an entire wireless network at the expense of the slightly
increased RTS/DRTS requests in the establishment process.
I. INT ROD UC TI ON
With the popularity of wireless local access, there has
developed an increasing demand for improving the throughput
and energy efficiency in data transmission between terminals
and access points/base stations. Traditionally, wireless net-
works are designed to provide single hop transmission, either
to Wireless Local Area Networks(WLAN) access points or
to base stations. However, the explosive implementation of
wireless networks has sparked the idea of Wireless Mesh Net-
works(WMN) [1], extending the wireless coverage, improving
the overall capacity and enabling network auto-configuration.
The wireless mesh network is a network consisting of com-
munication nodes organized in a mesh topology, which also is
a form of wireless ad hoc network. Wireless mesh networks
often consist of mesh clients, mesh routers and gateways [2].
The use of multi-beam antennas(smart antennas) in wireless
mesh networks has received growing attention thanks to the
higher antennas gain, better spatial reuse, longer transmission
range, as well as lower interference between multi-beam anten-
nas [3]. The increasing interest in WMN and their applications
in battlefield and disaster relief environment has later evolved
to a broader arena [4]. The research on WMN has attracted
both academe and industry, and has resulted in the rapid com-
mercialization as well as numerous standardization efforts [5].
With the multi-beam smart antenna system, multiple omni-
antenna or directional-antenna nodes may transfer data to or
from the other nodes simultaneously, increasing the potential
throughput substantially.
The omni-directional antenna usually spreads the elec-
tromagnetic energy of wireless signal over a large area in
space, while only a very small portion is actually received
by the intended receivers, thus limiting the overall capacity
and performance. Also, the omni-directional antenna usually
has the problems of multipath fading, delay spread and co-
channel interference(CCI) [6]. Currently, with the help of the
availability of low-cost computing capacity and the develop-
ment of new algorithms for processing signals from arrays of
simple antennas, beamforming antennas have become available
to wireless communication systems. Usually, the beamforming
antennas have arrays of simple smart antennas, which consist
of multi-beam antennas(MBA). The multi-beam antennas can
enhance the radiating electromagnetic waves in wireless com-
munications by actively controlling the temporal paces among
the radiating elements of the antenna array, using the Digital
Signal Processing(DSP) units.
A typical wireless local area network consists of the Access
Point(AP) and a finite set of mobile stations. Generally, the
AP is much more powerful and less physically constrained
than the mobile stations. The AP usually equips multiple smart
antennas to boost the network throughput by exploiting spatial
reuse [7]. The existing multi-beam smart antennas could be
broadly classified into three categories: switched multi-beam
antennas, adaptive array antennas, and multiple-input-multiple-
output(MIMO) links [8]. Each of these antenna technologies
has its pros and cons, which we will discuss them later.
The switched multi-beam antennas are relatively simple and
commercially available, and have been deployed in many real
applications [3].
Meanwhile, the superior capabilities of smart antennas can
be leveraged through appropriately designed upper layer net-
work protocols, including the Medium Access Control(MAC)
protocol. However, there exist several design challenges for
the MAC protocol compared with the traditional wired MAC
protocols. The traditional network protocols were originally
designed to run on the nodes equipped with omni-directional
antennas, and failed to interact with the underlying smart
multi-beam antennas. They may deteriorate the overall per-
formance even below the level achieved by omni-directional
antennas without the appropriate control [9]. Hence, it is
essential to investigate innovative protocols, specially in the
MAC layer, that are capable of harnessing the potential benefits
of smart multi-beam antennas in wireless mesh networks.
The rest of this paper is organized as follows: Section II
briefly describes the considerations of designing MAC proto-
cols. Section III presents the Multi-Beam Directional MAC
protocol. Section IV evaluates the performance and results.
Section V concludes this paper.
2017 International Conference on Computing, Networking and Communications (ICNC): Wireless Communications
978-1-5090-4588-4/17/$31.00 ©2017 IEEE
II. PROTOC OL DE SI GN CO NS ID ER ATIO NS
As we mentioned early in this paper, two main types of
multi-beam smart antenna systems have been widely studied
in the current literature: one is based on adaptive arrays and
the other is based on the fixed beam directional antennas. This
section will discuss the main challenges for MAC protocols in
a general beamforming environment.
A. Beam-synchronization constraint
Multi-beam antennas need the cooperation so that the AP
works correctly with other beams. To avoid the cosite inter-
ference problem, all sections at the AP must be in either the
transmission mode or the reception mode [3]. The scheduling
policies need to deal with the synchronization issue.
B. Hidden terminal problem
There exists the hidden terminal problem in the traditional
wireless network, occurring when two nodes are outside of
their carrier sensing ranges of each other during CSMA and
both of them attempt to communicate with the common node.
The solution for this traditional problem is by implementing
the RTS/CTS handshaking mechanism before data transmis-
sion to avoid the occurrence of collision.
In the multi-beamforming antennas, there exists a new
hidden terminal problem. When a potential interferer cannot
receive the RTS/CTS exchange information, due to its an-
tenna orientation during the handshake, and then initiates a
data transmission, a collision may occur. There are two new
types of hidden terminal problem [5] [10]: Hidden Terminal
Due to Asymmetry in Gain and Hidden Terminal Due to
Unheard RTS/CTS. The hidden terminal problems are much
more serious in a wireless mesh network, as it needs more
complex MAC protocols to synchronize the exchanged data
information among the neighbour nodes, especially for dealing
with mobility.
C. Deafness
Deafness may occur and is by far the most critical chal-
lenge in a wireless network, as first identified in the context of
the basic directional MAC protocol [14]. When a transmitter
tries to communicate with a receiver, however, this trying
transmission process fails because the receiver is beamformed
towards another direction that is away from the transmitter. The
consequences of deafness may be even worse and may lead
to short-term unfairness between flows that share a common
receiver if the involved transmitter has multiple packets to
send and constantly transmits the data by choosing a backoff
interval from the minimum contention windows. Moreover, the
deadlock may happen when a chain of deafness is possible in
which each station attempting to communicate with a deaf
station becomes itself deaf to another station [14].
D. Beam-overlapping
Multi-beamforming technique implements multiple beam-
formed beams in the smart antenna, when there is the interfer-
ence between different beams. Due to the physical imperfec-
tion of beamforming antennas, there generally exists a small
portion of beam-overlapping area for two adjacent beams. If
a station lies in the beam-overlapping area, this station can
hear data transmission from multiple sectors; in turn, multiple
sectors in this access point AP can also hear data transmission
from that station. This is not what the multi-beam wants.
Also, the beam-overlapping problem is much more serious in
a multipath rich environment. It will degrade omni-directional
communication and no spatial reuse can be exploited.
E. Unnecessary deferment
For multi-beam antennas, there exists two kinds of unnec-
essary deferment problems, the Head-of-Line(HoL) blocking
problem with beamforming MAC protocol, and one caused by
the rules of CSMA/CA.
Omni-directional antennas typically use First-In-First-
Out(FIFO) queueing policy to buffer the received signals and
this policy works fine in the omni-directional antenna because
it uses the same medium for all outstanding packets. When the
medium is busy, no packets can be transmitted. However, in the
case of multi-beamforming antennas, the medium is spatially
divided and it may be available in some directions but not
others. This is because the multiple beams are separated from
each other and have their own communication ranges. If the
packet at the top of the queue, using FIFO queueing policy, is
destined to a busy station or beam, then it will block all the
subsequent packets, even though some of them can be trans-
mitted. The HoL blocking problem maybe aggravated when the
top packet goes into a round of failed retransmissions including
their associated backoff periods [15]. Due to their geographical
proximity to each other in multi-beam, or even existing beam-
overlapping, one station can hear another station’s signal and
this may subdue the transmission of data.
F. Miss-hit
Most wireless communications are based on the rule of
Direction of Arrival(DoA) [16]. In the wireless mesh network,
stations tend to be more movable, thus challenging the wireless
communication. Under the basis of DoA estimation techniques,
when one station tries to send packets to the AP with a smart
antenna system, the AP can identify which beam(or which
beams if beam-overlapping problem or back/side-lobe problem
occur) is located in the sending station [3]. Due to the mobility
of a wireless mesh network, the beam location information
cached in the AP maybe stale and incorrect, and may need
to be updated according to a new location, when the mobile
stations move. In this case, the AP may direct a wrong beam
for the corresponding downlink transmission, leading to the
miss-hit problem.
III. THE MBDMAC PROTOCOL
There are two types of practical directional antennas:
phased-array antenna and switched-beam antenna. The phased-
array antenna achieves beam steering by constantly changing
the phase of the antenna elements that constitute the array;
the switched-beam antenna is equipped with a number of
directional antenna elements oriented in predefined directions.
The phased-array antenna is less cost effective than the
switched-beam antenna [17] [18]. In this paper, we propose
a Multi-Beam Directional MAC protocol(MBDMAC), using
the switched-beam antenna at each node that consists of M
beam-sectors.
2017 International Conference on Computing, Networking and Communications (ICNC): Wireless Communications
A. Mechanism Description
In this section, we present a Directional-to-Directional
MAC protocol for each communication. Assume that idle
nodes, or potential receivers, continuously scan through their
antenna sectors to emulate omni-directional antennas checking
for potential senders. When the nodes hear a Directional
RTS(DRTS) intended for themselves, they record and lock
in the respective direction and respond with a Directional
CTS(DCTS) after the evaluation processes.
To minimize the effects of the directional deafness, the
antenna continuously switches the sensing direction(beam-
sectors) of its idle node in a clockwise or anticlockwise
manner. Such behavior, essentially, emulates the presence of an
omni-directional antenna at the receiver. The time span that the
idle node spends in each beam-sector should be at least DRTS
+ SIFS + λBO , where λBO is the compensation time that a
sending node may spend in the back-off(BO) process.
To cooperate with each node and each sector, the network-
allocation vector(NAV) which is used in the IEEE 802.11 MAC
protocol is needed. As the IEEE 802.11 MAC protocol states,
each node maintains a NAV, updates from the duration field
of overheard RTS/CTS packages. In the directional antenna
scenario, the similar DNAV mechanism proposed in [19],
which is different from that of IEEE 802.11 protocol is used.
For efficiency of the MAC protocol, the senders need to
estimate the potential directions of the intended receiving node,
and each node estimates and caches the AoA(Angel-of-Arrival)
of any message it hears or overhears. Before sensing the
medium, the sending node checks its AoA cache to determine
the receiver’s most likely directions. When a node hears a new
message from its neighbors, the AoA information is updated.
Also, each AoA entity has an expiration time.
When a node hears the transmission on one of its sectors,
it sets its corresponding DNAV and continues to sequen-
tially scan through all other sectors. This can reduce the
deafness problem and alleviate the directional hidden-terminal
problems, however, it comes with an increased number of
handshaking messages that need to be sent. The number of
handshaking messages partly depends on the back-off algo-
rithm and the number of antenna sectors in the antenna.
In our protocol, we also assume that, in addition to the
Mbeam-sectors, a single dedicated channel for each sector
is available to the multi-beam antenna and this channel will
be used as a control channel(CCH). The control channel is
supported by the RF-chain so that communication in the
CCH can take place simultaneously with data communication.
The MBDMAC protocol consists of three phases: connection
initialization, channel contention and data communication.
B. Connection Initialization
The basic idea of an MBDMAC protocol is to introduce a
timing structure to facilitate multiple handshakes(sequential or
overlapping) before parallel collision-free data transmissions.
We use the Subscriber Station(SS) and Base Station(BS) as
the models to illustrate the connection initialization process.
The connection initialization phase is based on control
information exchanged through the out-of-band CCH. Upon
switching on, the SS sends an omni-directional Quiet Period
Request(QPREQ) message by CCH. The coordinate infor-
mation of SS are contained in a QPREQ message, which
the receiving BS stores into its the geo-location databases.
The BS that receives this request sends back a Quiet Period
Reply(QPREP) indicating the current time, the starting time
of next QP and the QP duration and repetition interval. This
message also contains the locations of all other BSs that
are within the communication range of the requesting SS.
QPREPs are sent directionally after a random back-off time.
The receiving SS first listens to the medium omni-directionally.
Once the SS detects a transmission, it switches to directional
reception to limit QPREP collisions.
Each sector maintains a DNAV which contains the channel
information covered by the beam of the BS. Both BS and SS
use their own DNAVs to flag which sectors are more suitable
for communication during the channel contention phase. The
BS keeps track of the data channel in each beam and the
cumulative, in all beams, channel utilization. Meanwhile, the
SS also keeps track of the availability of BS in each beam.
C. Channel Contention
The contention resolution scheme facilitates multiple nodes
to win out, and the winning nodes will be collision-free with
each other. Collision-free means that the winning nodes will
not collide with each other when they simultaneously send
DATA to or receive ACK from the access point during the
period of Data Communication. In order to be collision-free,
different winning nodes must be in the different beam-sectors.
To fully facilitate the capacity of a multi-beam antenna, we
use the parallel transmission, which means the antenna can
transmit the ACK and DATA in parallel for each sector. In the
phase of DATA transmission, multiple back-to-back packets
are allowed to be transmitted by one node as long as the total
time for transmitting these multiple packets does not exceed
the length of collision-free parallel data transmission. Note that
power control and/or auto rate schemes can be incorporated in
this scheme to further increase the throughput and improve the
energy efficiency. In the period of parallel ACKs, the access
point SS will transmit one ACK for each user concurrently
because the users are in different sectors.
The channel contention scheme is based on CSMA/CA.
However, unlike the traditional CSAM with an omni-
directional access point, two nodes, simultaneously transmit-
ting RTSs to the access point, may not collide with each other
if they aim to different beam-sectors. In other words, if there
is only one node sending RTS in a given beam-sector and this
node is free from the beam-overlapping problem and multipath
problem, the node succeeds no matter how many nodes are
concurrently sending RTSs in other beam-sectors. Unless each
beam-sector has collisions of RTSs, the access point will send
CTSs in all beam-sectors concurrently after receiving RTSs.
After receiving these RTSs, the SS decides to connect with
which a sector of BS based on the matching assigned/available
channels, link quality and channel utilization. Developing an
efficient algorithm for selecting the most suitable sectors of
BS is out of the scope of this paper. After the beam sector
selection process, the SS informs the selected sector of BS
on the channel for connection with a directional Request to
Connect(RTC) message, then the BS acknowledges with a
2017 International Conference on Computing, Networking and Communications (ICNC): Wireless Communications
Fig. 1. Conceptual Control Flow for MBDMAC Protocol
directional Clear to Connect(CTC) message. The RTC also
includes all available channel information of the corresponding
SS in its beam. The RTC messages are sent using CSMA/CA
and the process is repeated if no reply is received.
D. Data Communication
The data communication in the MBDMAC protocol is
very similar to the DMAC/DA protocol [21]. During the data
communication process, only the beam-sector of BS can send
a Wait to Send(WTS) message. There is no point in SS sending
WTS messages since at any point in time an SS is assigned
to and can receive data from at most one sector of BS. WTS
messages should be sent to active nodes in all assigned data
channels. All other messages in the DRTS-DCTS-WTS-Data-
ACK processes are sent in the same data channel. However,
the purpose of the WTS message is not to prevent collision but
to mitigate the deafness problem by warning wireless nodes
not to send DRTSs that will not be received and will result in
unnecessary back-offs and resource consumption.
When the SS node is idle, the SS senses only the data
channel in which it communicates with its associated sector
of BS. However, the BS needs to be able to sense many data
channels simultaneously for DRTS.
E. Control Flow
In our schemes, multi-beam antenna keeps checking
whether there are any of its RTSs in idle states. If the antenna
listens to one of its RTSs intended to itself, it checks its own
DNAV table for the corresponding beam-sector. If the request
is not in the DNAV table, this request is added, otherwise
it is updated with this new value. Once the current beam-
sector of the antenna finishes its sensing, the antenna checks
whether all Msectors are finished or not; if not, it goes to the
next beam-sector, otherwise, this round sensing is finished. The
antenna with all its DNAV records in different beam-sectors
uses the machine learning algorithm to assign the directions
in different sectors to each potential sender. How to get an
optimal assignment for each potential connection is out of this
scope. It must consider a lot of parameters and scheduling
issues [20], such as the distance between communication
nodes, energy consumption, and network quality. The control
flow is shown in Fig. 1.
IV. PER FO RM AN CE EVAL UATION
To evaluate the performance of the proposed MBDMAC,
we developed a custom event-driven simulator in C language.
In this section, we will present the simulation parameters used
in the experiment and the corresponding simulation results.
Fig. 2. Ideal Case for Different Number of Sectors
A. Simulation Parameters
Suppose the system has Mdisjoint beam-sectors, each of
which has npotential users. The access point always has
packets in the queue for transmission to each node, and each
node always has packets for transmission to the access point.
Our simulation is based on the Network Simulator(ns-2) [22].
Some of the values of the parameters used in the simulation
are listed in Table I. The source-destination pairs are one-hop
neighbors and each pair is loaded with a constant-bit-rate flow.
Each simulation runs for at least 200 seconds.
TABLE I. SI MUL ATIO N PARA MET ER S
Parameter Value
Basic Rate 2 Mbps
PHY header 192 bits
MAC header 28 Bytes
Package size 1000 bytes
SIFS 10 µs
DIFS 40 µs
RTS 168 bits
CTS 136 bits
ACK 112 bits
Wmax 64
B. Simulation Results
We first examine the ideal saturated case, which can
provide a performance limit we can compare. Here we assume
the beam-forming is perfect and there are no beam-overlapping
problems or multi-path related problems. Fig. 2 compares the
throughput obtained from the simulation. For the case of M=
1, the original IEEE 802.11 DCF MAC protocol is used; for
M > 1, the proposed MBDMAC protocol is used, where M
represents the number of beam-sectors. Also, here we select a
relatively small value of Wmax = 64 as the maximal contention
window size at the receiver side. As shown in Fig. 2, the
proposed scheme can achieve about 1.61, 2.30 and 3.03 times
of the case of IEEE 802.11 throughput on average, when the
beam-sector number Mis 2, 3 and 4, respectively.
Also, we simulate the network throughput for different
Wmax with respect to the different number of antenna sectors,
as shown in Fig. 3. The Wmax is chosen to be 64, 128 and
256, respectively, and the number of beam-sectors is increased
2017 International Conference on Computing, Networking and Communications (ICNC): Wireless Communications
Fig. 3. The Relation of Network Throughput and the Number of Sectors
Fig. 4. The Number of RTS/DRTS with respect to the Time Span
from 2 to 4 and then to 6. From the figure, we can see
that when the number of sectors is increased from 2 to 4,
the network throughput is continually increased. However,
when the number of antenna sectors is further increased, the
throughput is lower than the case when 4 antennas sectors
are used. This is mainly due to the increased overhead of
control messages, such as RTS or DRTS. Furthermore, this
phenomenon leads to the conclusion that there is an optimal
combinational number of antennas sectors and Wmax for a
specific network density.
To better understand the network throughput, we further
count the number of RTS and DRTS generated during the
simulation, with respect to the number of beam-sectors. Fig. 4
shows the total number of DRTSs sent, including multiple
DRTS attempts for the same data packet, as compared to
the number of RTSs sent in the case of omni-directional
communication.
V. CO NC LU SI ON
In this paper, a novel MAC protocol for multi-beam di-
rectional antennas has been proposed. This protocol is partly
based on the CSMA/CA scheme and a dedicated control
channel with DNAV table. Protocol, to detect the optimal sec-
tors/channel for data connection in a multi-beam environment,
the decision to connect is delayed until all beam-sectors in the
antenna have finished sensing. Simulation results show that it
significantly improves the wireless throughput.
REF ER EN CE S
[1] Akyildiz, Ian F., Xudong Wang, and Weilin Wang. ”Wireless mesh
networks: a survey.” Computer networks 47.4 (2005): 445-487.
[2] Akyildiz, Ian, and Xudong Wang. Wireless mesh networks. Vol. 3. John
Wiley & Sons, 2009.
[3] Wang, Jianfeng, Yuguang Fang, and Dapeng Wu. ”Enhancing the per-
formance of medium access control for WLANs with multi-beam access
point.” Wireless Communications, IEEE Transactions on 6.2 (2007): 556-
565.
[4] Conti, Marco, and Silvia Giordano. ”Multihop ad hoc networking: The
reality.” Communications Magazine, IEEE 45.4 (2007): 88-95.
[5] Bazan, Osama, and Muhammad Jaseemuddin. ”A survey on MAC
protocols for wireless adhoc networks with beamforming antennas.”
Communications Surveys & Tutorials, IEEE 14.2 (2012): 216-239.
[6] Winters, Jack H. ”Smart antennas for wireless systems.” Personal Com-
munications, IEEE 5.1 (1998): 23-27.
[7] Chou, Z., C. Huang, and J. Chang. ”QoS Provisioning for Wireless LANs
with Multi-beam Access Point.” (2013): 1-1.
[8] Sundaresan, Karthikeyan, and Raghupathy Sivakumar. ”A unified MAC
layer framework for ad-hoc networks with smart antennas.” Proceedings
of the 5th ACM international symposium on Mobile ad hoc networking
and computing. ACM, 2004.
[9] ATHAN, RAMRAMAN. ”Antenna beamforming and power control for
ad hoc networks.” Mobile ad hoc networking (2004): 139.
[10] Choudhury, Romit Roy, et al. ”On designing MAC protocols for
wireless networks using directional antennas.” Mobile Computing, IEEE
Transactions on 5.5 (2006): 477-491.
[11] Choudhury, Romit Roy, et al. ”Using directional antennas for medium
access control in ad hoc networks.” Proceedings of the 8th annual
international conference on Mobile computing and networking. ACM,
2002.
[12] Gossain, Hrishikesh, et al. ”The deafness problems and solutions in
wireless ad hoc networks using directional antennas.” Global Telecom-
munications Conference Workshops, 2004. GlobeCom Workshops 2004.
IEEE. IEEE, 2004.
[13] Takata, Masanori, Masaki Bandai, and Takashi Watanabe. ”A MAC
protocol with directional antennas for deafness avoidance in ad hoc
networks.” Global Telecommunications Conference, 2007. GLOBE-
COM’07. IEEE. IEEE, 2007.
[14] Choudhury, Romit Roy, and N. F. Vaidya. ”Deafness: A MAC problem
in ad hoc networks when using directional antennas.” Network Protocols,
2004. ICNP 2004. Proceedings of the 12th IEEE International Conference
on. IEEE, 2004.
[15] Bazan, Osama, and Muhammad Jaseemuddin. ”An opportunistic di-
rectional MAC protocol for multihop wireless networks with switched
beam directional antennas.” Communications, 2008. ICC’08. IEEE In-
ternational Conference on. IEEE, 2008.
[16] Stevanovic, Ivica, Anja Skrivervik, and Juan R. Mosig. ”Smart antenna
systems for mobile communications.” Ecole Polytechnique Federale De
Lausanne, Technical Report (2003).
[17] Gang Wang, Wenming Li, and Xingyu Cai. “MBDMAC: A MAC
Protocol for Multi-Beam Directional Antennas in Wireless Networks.”
Mobile Ad hoc and Sensor Systems(MASS), 2015 12th International
Conference on. IEEE, 2015.
[18] Shihab, Emad, Lin Cai, and Jianping Pan. ”A distributed asynchronous
directional-to-directional MAC protocol for wireless ad hoc networks.”
Vehicular Technology, IEEE Transactions on 58.9 (2009): 5124-5134.
[19] Choudhury, Romit Roy, et al. ”Using directional antennas for medium
access control in ad hoc networks.” Proceedings of the 8th annual
international conference on Mobile computing and networking. ACM,
2002.
[20] Wang, Gang, Wenming Li, Xiali Hei. ”Energy-Aware Real-time
Scheduling on Heterogeneous Multi-Processor.” 49th Conference on
Information Sciences and Systems, 2015, Baltimore, Maryland.
[21] Takata, Masanori, Masaki Bandai, and Takashi Watanabe. ”A MAC
protocol with directional antennas for deafness avoidance in ad hoc
networks.” Global Telecommunications Conference, 2007. GLOBE-
COM’07. IEEE. IEEE, 2007.
[22] http://www.isi.edu/nsnam/ns/
2017 International Conference on Computing, Networking and Communications (ICNC): Wireless Communications