JOURNAL OF COMMUNICATIONS SOFTWARE AND SYSTEMS, VOL. 15, NO.

2, JUNE 2019 109

A Hybrid based Distributed Slot Scheduling

Approach for WSN MAC
Manas Ranjan Lenka, Amulya Ratna Swain and Biraja Prasad Nayak

Original scientific paper

Abstract—In Wireless Sensor Networks(WSNs), collision han- has been sent from other nodes to it then that situation is
dling during transmission of data is an important challenge. MAC referred as idle listening. This is one of the major source of
protocol plays a vital role in handling those collisions. Among energy waste in WSN. Various protocols [1], [2] have been
different types of MAC protocols, schedule based MAC protocol
is one where a valid schedule is prepared to handle the collision. designed to reduce idle listening by switching off the radio
The existing schedule based MAC protocols focus on preparing from time to time.
either a feasible schedule or an optimal schedule. In order to Overhearing: Another source of energy waste is overhearing.
satisfy both feasibility as well as optimality feature, in this paper, In the case of overhearing, a node receives data which is not
we proposed a hybrid approach for slot scheduling that prepares meant for it there by wasting energy.
a feasible schedule in a distributed manner and at the same
time reduces the number of slots in the feasible schedule to Control message overhead: In WSN, it is unavoidable to stop
achieve optimality. In this paper, we named this as Hybrid based the exchange of control messages among sensor nodes before
Distributed Slot Scheduling (HDSS) approach. The proposed the actual communication. The exchange of control messages
HDSS algorithm initially prepares a feasible schedule which is also consumes energy. Hence in order to conserve energy,
further tuned in quick time to prepare a valid schedule with a WSN protocols must exchange minimum number of control
reduced number of slots. The reduction of the number of slots in
the schedule improves the efficiency of data transmission in terms messages before the actual data communication.
of latency. The simulation results show that the HDSS algorithm Collision: During data communication, if a data packet gets
outperforms RD-TDMA with respect to both the number of slots collided with another packet then the transmission of the data
allotted for a feasible schedule as well as the data transmission packet must be stopped and retransmission of the same packet
latency. must be carried out which consumes energy. Also, due to the
Index Terms—Wireless Sensor Network, Media Access Control, collision of packets, the latency in the transmission of the
Slot Scheduling, feasible schedule, correlated contention. packets has been introduced.
Among various sources of energy wastage, reducing col-
I. I NTRODUCTION lision during the data transmission is one of the impor-
tant consideration to conserve energy in WSN. Various slot
W SNs consist of thousands of tiny sensor nodes that
sense the environment, collect the data, process it, and
send the information to the sink node. The sink node behaves
scheduling algorithms [3], [4], [5], [6], [7], [8], [9], [10],
[11], [12], [13], [24] and MAC protocols [1], [2], [14], [15],
as a gateway which provides the information to the outer [16], [17], [18], [19], [20], [25], [26] has been proposed
world. In this current age of advancement, WSN plays a vital to fulfill the above requirements. Nevertheless, none of the
role in various application areas such as Health Monitoring, existing literature focuses on achieving both feasible schedule
Habitat Monitoring, Home automation, Environmental Mon- and optimal schedule in quick time. So, in order to achieve
itoring, Military Surveillance, Bridges and building Monitor- both feasibility as well as optimality, here, we proposed a
ing, and Real-time Monitoring of Nuclear Plants. Each sensor distributed approach for hybrid slot scheduling algorithm that
node in WSN is powered by a battery, and these are mostly not only prepares a feasible schedule but also further reduce
deployed in the hostile environment therefore not possible to the number of slots along with handling collision.
recharge. Hence in WSNs, one of the most important goals is The remaining part of the paper is structured as follows.
to conserve the energy as the battery power is limited. Along Section II describes the related works. The proposed algorithm
with limited battery power, some of the other constraints in is briefly illustrated in Section III. Section IV analyzes the
WSNs are processing capability, memory, and bandwidth. proposed algorithm. The simulation results and its analysis is
In order to conserve energy it is indeed required to identify presented in Section V. The conclusion and future work is
various sources of energy wastes and handle them properly.The given in Section VI.
various sources of energy wastes in WSNs are basically
categorized into four types [1], [2]. II. R ELATED W ORK
Idle listening: Any node that has kept its radio on but no data
The MAC protocols for WSNs can be broadly classified into
Manuscript received January 16, 2019; revised April 9, 2019. Date of 3 categories [21] viz., Schedule-based, Contention-based, and
publication May 16, 2019. Date of current version June 3, 2019. Hybrid. The Schedule-based MAC protocols prepare a valid
Authors are with the KIIT University, Bhubaneswar, India (e-mails:
{manasy2k3, swainamulya, brjnayak}@gmail.com). schedule for data communication to handle the collision, which
Digital Object Identifier (DOI): 10.24138/jcomss.v15i2.695 is one of the prominent source of energy wastes in WSNs [1].

1845-6421/06/695 © 2019 CCIS

110 JOURNAL OF COMMUNICATIONS SOFTWARE AND SYSTEMS, VOL. 15, NO. 2, JUNE 2019

Although this category of MAC protocols handle the collision Dam et al. [2] proposed the T-MAC protocol which is
efficiently, still it does not adapt well to the topology changes a contention based MAC protocol and works very much
in WSN. The Contention-based MAC protocol adjusts to the alike the S-MAC protocol. The T-MAC protocol improves the
changes in topology (i.e. addition or removal of any sensor performance of S-MAC by adapting a variable duty cycle in
nodes to WSN) very easily. Nevertheless, handling collision a novel way to cope up with the variable load condition.
in this category of protocols is costlier as compared to the Z-MAC(Zebra MAC) proposed by Rhee et al. [14], is a
Schedule-based MAC protocols. Furthermore, the hybrid MAC hybrid MAC protocol which combines both the advantages of
protocols are those protocols that try to combine the advan- CSMA and TDMA to enhance the channel utilization. In the
tages of both schedule based MAC protocol and contention worst case, the performance of Z-MAC falls back to CSMA.
based MAC protocol while counteracting their weaknesses. This protocol works in 2 phases as follows.
In many sensor network applications, the sensor nodes Set-up phase: During the set-up phase, first 2-hop neighbor list
remain idle for a long time which consumes a lot of energy. of each sensor node is prepared. Then each node is assigned
Ye et al. [1] proposed a contention based MAC protocol called a time slot using DRAND [22] in such a way that no 2-
S-MAC that minimizes the wastage of energy due to idle hop neighbors get the same slot number. The node which is
listening and also taking care of other forms of energy wastage assigned to a slot is known as the owner of the slot and all
due to collision, overhearing, and control message overhead. other nodes are the non-owners of that slot.
According to S-MAC protocol, each sensor node goes to sleep Transmission phase: In the transmission phase, a node can
state from time to time by turning off its radio. After timeout, transmit data in any time slot. Before transmitting the data,
the sensor node wakes up and listens if any neighbor nodes the node first sense the carrier using CSMA and if it finds the
wants to send data or the same node has some data to sent to carrier to be free, then transmits the data using that time slot.
its neighbors. In order to achieve the above mentioned benefits, However, the owner of that slot always gets the priority over
the following steps are carried out. the non-owners for sending the data.
Periodic listen and sleep: Before each node goes to periodic Slama et al. [15] proposed a hybrid MAC protocol called
listen and sleep, they must synchronize with each other by I-MAC that takes the advantage of both CSMA and TDMA
broadcasting their schedule to its Neighbors. After synchro- technique with an adaptive priority scheme for channel access.
nization, each node has a schedule table which keeps the Like Z-MAC, this protocol also follows 2 phases, i.e. setup
schedules of all its neighbors. phase and transmission phase. During the setup phase, it per-
The below steps are followed to prepare and synchronize forms neighbors discovery, TDMA slot assignment, framing
the schedule among the neighbor nodes. and synchronization operations only once. These steps are also
• During start up, a sensor node first listens for certain carried out when there is a significant change in the network
amount of time to receive a schedule from it’s neighbors. topology.
If it does not hear anything from it’s neighbors within The TDMA slot scheduling is done using the proposed
the timeout period, then it selects its own schedule and distributed neighborhood information based algorithm (DNIB)
immediately broadcasts the same to its neighbors. [23] that prepares the schedule based on 2-hop neighbors and
• After receiving this schedule, the sensor nodes use the handles the dynamic topology changes with low complexity.
same without creating their own schedule and then, they In the data transmission phase, the owner nodes always get
wait for a random amount of time and broadcasts the priority over the non-owner nodes for data transmission. In
received schedule to their neighbors. case, the owner nodes have no data to sent at a particular
• The sensor nodes, who receive a different schedule other point of time then the non-owner nodes can use that slot for
than their own schedule will follow both the schedules data transmission. However, the non-owner nodes with higher
but broadcasts only their own schedule before going to priority compete with one another to use that slot for data
sleep. transmission. Hence, the chance of collision is reduced as the
Collision avoidance: After synchronization, nodes whoever try number of competing nodes are less. The nodes having more
to transmit data, first sense the carrier and exchanges RTS/CTS data to transmit are assigned with higher priority so as to
message to avoid a collision before any data communication. utilize the channel in a better way.
Overhearing avoidance: The neighbor sensor nodes of both Zhuo et al. [16] proposed the Queue-MAC protocol which
the sender and the receiver of an RTS/CTS packet are brought maintains a queue of nodes those who are willing to transmit
to the sleep state to avoid overhearing. data. This protocol dynamically adapts the duty cycle using
S-MAC [1] protocol makes a trade-off between the energy the queue length of the nodes. When the queue length gets
consumption and throughput/latency. The throughput is re- increased i.e. traffic increases, the protocol extends the active
duced as the data transmission occurs only during the active CSMA period by adding dynamic TDMA time slots. Hence,
period. In addition the latency for data transmission may at low load it preserves energy and at the same time at high
increase, as there is a chance that an event might occur during load, it provides high throughput.
the sensor node is in sleep mode and data transmission has This adaptive CSMA/TDMA hybrid MAC is an improve-
to be deferred till the sensor node has been brought back to ment over the IEEE 802.15.4 protocol. IEEE 802.15.4 is a
the active mode. S-MAC employs a fixed sleep/active period standard for Low-rate wireless personal area networks (LR-
which also affects the performance of the protocol in varying WPANs). The IEEE 802.15.4 standard operates in 2 modes
load condition. i.e. beacon enabled mode and non-beacon mode.
M. R. LENKA et al.: ARDUINO-BASED SOLUTION FOR IN-CAR-ABANDONED INFANTS' CONTROLLING REMOTELY 111

In non-beacon mode, CSMA/CA is used as the MAC layer • The whole time line is divided into a number of fixed
protocol. In beacon enabled mode, a special node called sized frames and each frame is consisting of no. of fixed
the PAN coordinator sends beacon frames in a regular time time slots (i.e. the maximum of all the 2-hop neighbors)
interval (i.e. beacon interval) to identify its PAN. The beacon , under the assumption that nodes are not synchronized
interval consists of 2 periods with reference to a global clock.
• Contention Access Period (CAP): During this period • Initially, each node tries to occupy a slot by broadcasting
nodes get access of the channel using slotted CSMA/CA a request message to its one-hop neighbor.
technique. • When a node receives this request message, either it sends
• Contention Free Period (CFP): During this period the a grant or reject message to the sender.
nodes required to access the channel, and sent a request • It sends a grant message to the sender if the node itself
to the coordinator which gives a time slot to the node. has not requested for the same slot or it has not granted
the same slot to any other node. Otherwise, send a reject
In this protocol, during the Contention Access Period message.
(CAP), nodes can access the channel using an adaptive • When a node requests for a particular slot and it is granted
CSMA/TDMA hybrid MAC protocol rather than slotted by all of its neighbors through the grant message then the
CSMA/CA protocol. The channel utilization and amount of requested node is assigned to that slot. In case, it receives
data pending in the queue of each node decides the border be- a reject message from any of its one-hop neighbor nodes
tween CSMA/CA and TDMA protocol during the Contention then it stops there and reiterates the same process with
Access Period(CAP). The coordinator checks the queue state another available slot.
of each node and assigns a TDMA slot to the node in the • Once a node is allotted to a slot then it broadcasts the
descending order of their amount of pending data in the queue. node allotment information to its two-hop neighbors from
Once a TDMA slot is assigned to a node for communication, it time to time. This helps the two-hop neighbors of the
is no more going to participate in the contention for getting the node to refrain from either request for getting that slot or
channel using CSMA/CA and thereby reduces the number of grant that slot to others.
nodes contending for the channel. Hence, reduces the collision
This paper is mainly focusing on preparing both feasible
and improves the performance in terms of energy consumption
as well as optimal schedule, which will handle the correlated
and throughput.
contention as well as minimize the number of slots to reduce
Rhee et al.[22] proposed a randomized distributed TDMA the latency. In order to do so, our proposed algorithm initially
slot scheduling algorithm called DRAND. In this algorithm, prepares a feasible schedule and then fine tuned the feasible
each node can be either in one of the four states i.e. IDLE, schedule in a novel way that reduces the number of slots to
REQUEST, GRANT, or RELEASE. Initially, each node is achieve the desired objective.
in the IDLE state. Then each node goes for a lottery, and
whichever node wins the lottery that node goes into the
REQUEST state and gets the chance to negotiate with their III. P ROPOSED HDSS A LGORITHM
neighbors to choose a time slot. If a node is either in IDLE The proposed HDSS algorithm initially prepares a feasible
or RELEASE state and receives a request message from its schedule to avoid the collision. Once a feasible schedule gets
neighbors then the receiving node sends back a grant message prepared, it is further fine tuned in quick time to minimize the
and itself goes to GRANT state. While a node receives a number allotted slots with maintaining the feasibility of the
request message being in the REQUEST or GRANT state newly prepared schedule intact. In order to do so, the proposed
then the receiver sends back a reject message. After receiving HDSS algorithm goes through following three phases.
the reject message the sender node goes back to IDLE state.
When a node, who is in the process of choosing a time slot, • Calculate the maximum two-hop neighbors count at each
receives grant message from all of its neighbors then it enters node by exchanging the number of two-hop neighbor
into RELEASE state and then broadcasts the release message among each other.
containing the information regarding the chosen slot. • Allotment of slot to each sensor node in such a way that
no two-hop neighbors are alloted the same slot thereby
Bhatia et al. [4] proposed an RD-TDMA scheduling algo-
handling the collision during data transmission.
rithm for handling collision in WSNs. Most of the existing
• Reduce total number of slots in a quick time and also
TDMA scheduling algorithm tries to find out an optimal sched-
handle the collision.
ule. However, this scheduling algorithm prepares a feasible
schedule to handle the correlated contention. When two or Our proposed algorithm use the following assumptions to
more nodes transmit data at the same time and collision occurs work correctly.
at any of the receivers, then this situation is referred as a • All the sensor nodes are static and homogeneous in
correlated contention. In the case of correlated contention, the nature.
contention occurs for a very short duration. Hence, RD-TDMA • Uniform random deployment of sensor nodes.
algorithm prepares a feasible schedule within a short period • Local clocks of all the sensor nodes are synchronized.
of time, which helps in handling the correlated contention • Packet transmission happens under ideal condition i.e.
in a better way. The RD-TDMA algorithm proceeds in the during packet transmission, there is no loss or corruption
following way. of packets.
112 JOURNAL OF COMMUNICATIONS SOFTWARE AND SYSTEMS, VOL. 15, NO. 2, JUNE 2019

The set of common notations used in our algorithm is

summarizes in Table I. Table II summarizes the set of notations
used for various message exchanges and Table III summarizes
various data structures maintained at each node.

TABLE I
S ET OF C OMMON N OTATIONS USED

Notation Description
SAi Original slot number allotted to the ith node
SRi Reallotted slot number to the ith node
N Total number of nodes
Nsq unit Number of nodes per square meter
Atot Total deployment area in square meter
STA Total number of slots allotted for a feasible schedule
NMax Maximum 2-hop neighbors count
NAvg Average 2-hop neighbors count Node Sending one hop ND Message One hop ND Message
SRA Reduced number of allotted slots as per the ratio One hop ND response
Node Sending one hop ND response Message
of average 2-hop neighbors count to maximum Message
2-hop neighbors count
SFA Total number of slots allotted finally after the reallotment
Fig. 1. One-hop neighbor discovery
process
O(i,j) ith node is the owner node of jth slot
N(i,j) ith node is the non-owner node of jth slot
RSi Re-allotment slot information at ith node [Step 4:] After discovering the list of one-hop neighbors, i.e.
N1i , a two-hop neighbor-discovery request (ND2Req ) message
that consists of the list of one-hop neighbor’s for node i.
TABLE II [Step 5:] Each node j that receives the two-hop neighbor
S ET OF N OTATIONS USED FOR VARIOUS M ESSAGE E XCHANGES discovery request (ND2Req ) message populates its two-hop
Notation Description
neighbor list (N2j ) using the one-hop neighbor list (N1i ) of
ND1Req One-hop neighbor discovery request message node i.
ND1Res One-hop neighbor discovery response message [Step 6:] After few such iterations each node i populates
ND2Req Two-hop neighbor discovery request message
NC2Max Max two-hop neighbor count message
their final two-hop neighbors list(N2i ) and then broadcasts a
SAReq Slot allotment request message NC2Max message.
SGRes Slot grant response message [Step 7:] Each node i that receives the NC2Max message
SRRes Slot reject response message
SASucc Slot allotment success message
verifies whether the received maximum two-hop neighbors
count (NMax ) is more than the maximum two-hop neighbors
count (NMax ) currently held at its own end. If so, then NMax
is updated at node i with the received maximum two-hop
TABLE III neighbors count. Then node i broadcasts the NC2Max message
S ET OF I NFORMATION M AINTAINED AT E ACH N ODE
over the whole network. Along with this, the node also
Notation Description calculates the average two-hop neighbors count(NAvg ) with
N1i List of one-hop neighbors of ith node the help of received maximum two-hop neighbors count(NMax )
N2i List of two-hop neighbors of ith node from a particular node. During the calculation of average two-
N2Max Maximum two-hop neighbors count of the whole network
ASi list of available slots at node i hop neighbors count(NAvg ), a node rejects all the duplicate
SS(i,j) Status of slot j at node i maximum two-hop neighbors count (NMax ) of a particular node
received from its different neighbors. Initially, the feasible
schedule is prepared with requisite number of slots using NMax ,
where NAvg helps in further minimizing the number of slots
A. Phase1: Two-hop Neighbor Discovery in the feasible schedule.
In this phase, we follow the steps as given below to calculate As per figure 1, the node that represented with black color
the two-hop neighbors count at each sensor nodes. The two- broadcasts an ND1Req message to find its one-hop neighbors.
hop neighbors count further helps in preparing the feasible The cross marked nodes (who receives this ND1Req message)
schedule. send back a ND1Res message to the black color node. Hence,
[Step 1:] Initially each sensor node i broadcasts an one-hop the cross marked nodes become the one-hop neighbors of the
neighbor discovery request (ND1Req ) message. black color node.
[Step 2:] Nodes that receives ND1Req message generate an As per figure 2, after the cross marked nodes find their
one-hop neighbor discovery response (ND1Res ) message and one-hop neighbors, the dotted nodes become the one-hop
send it back to the ith node. neighbors of the cross marked nodes. Then, the black color
[Step 3:] The nodes from which the node i has received node broadcasts a ND2Req message to their neighbors for the
the ND1Res message are considered as the list of one-hop discovery of the two-hop neighbors. Each cross marked nodes
neighbors (N1i ) of node i. sends a ND2Res message that contains the list of its one-hop
M. R. LENKA et al.: ARDUINO-BASED SOLUTION FOR IN-CAR-ABANDONED INFANTS' CONTROLLING REMOTELY 113

1 2 3 4 5 6 7 8 9

1 2 3 4 5 6 7 8 9
1 2 3 4 5 6 7 8 9

1 1 2 3 4 5 6 7 8 9

1 2 3 4 5 6 7 8 9

1 2 3 4 5 6 7 8 9
1 2 3 4 5 6 7 8 9

Node Sending slot allotment message Granted slot

One hop neighbors of black color node Requested slot

Node Sending two hop ND Message One hop ND Message
One hop ND response Message Slot allotment message Un-allotted slot
One hop-neighbors of black color Node Slot grant message
Two hop ND Message
Slot allotment success message Allotted slot
One hop-neighbors of Nodes with cross Two hop ND response Message

Fig. 2. Two-hop neighbor discovery Fig. 3. Success in slot allotment for nodes

neighbors. Finally, the cross marked nodes and dotted nodes

form the two-hop neighbors of the black color node. 1 2 3 4 5 6 7 8 9

1 2 3 4 5 6 7 8 9
B. Phase2: Allotment of Slots 1 2 3 4 5 6 7 8 9

After discovery of two-hop neighbors and N2Max information, 1 1 2 3 4 5 6 7 8 9

each node i prepares a feasible schedule with number of slots

(STA ) equals to the N2Max . During the slot allotment process 1 2 3 4 5 6 7 8 9
each slot is associated with one of the states represented in 1 2 3 4 5 6 7 8 9
Table IV. 1 2 3 4 5 6 7 8 9

TABLE IV Node Sending slot allotment message Granted slot

VARIOUS S TATE N OTATIONS OF E ACH S LOT Requested slot
One hop neighbors of black color node
Notation Description Slot allotment message Un-allotted slot
SU Slot is in Un-allotted state Slot grant message
Slot allotment failure message Allotted slot
SR Slot is in Requested state
SG Slot is in Grant state
SA Slot is in Allotted state Fig. 4. Failure in slot allotment for nodes

The details of the slot allotment process proceeds as follows. neighbors then the state of the requested slot is updated to
[Step 1:] Every slot in a feasible schedule is assigned with a allotted (SA) and broadcasts a SASucc message to its neighbors
state “un-allotted” (SU). or else, the state of the slot is updated to un-allotted(SU) and
[Step 2:] Each node i selects a slot randomly from the then go back to step-2 again.
currently available “un-allotted” list. Then the selected slot [Step 6:] Each node i that receives the SASucc message update
is updated with state as “requested” (SR). Finally, node i state of the slot to be allotted(SA). In case, node i did not
sends a slot allotment request message (SAReq ) that carries receive the slot allotment success message (SASucc ) within a
the requested slot information to its neighbors. certain period of time, after giving a grant to a particular slot,
[Step 3:] Each node that receives the SAReq message grant then node i update the slot status to be ”un-allotted” (SU) at
the requested slot if either of the following criteria matches or its own end considering the slot allotment fails.
else rejects the request for slot allotment. As per figure 3, the node represented with black color
• The receiver node itself has not requested for the same chooses a slot randomly and broadcasts a SAReq message to
slot earlier. its neighbors. Once the black color node receives the SGRes
• The receiver node has not yet granted the same slot to message from all of its neighbors, it allots the requested slot
any other node. for itself and inform the same to its neighbors through SASucc
[Step 4:] In case the slot is granted then the slot status is message. The same requested slot will be updated at the
updated to granted(SG) and a SGRes message is sent to its neighbors end to state SA when the neighbor node receives
neighbors. The intended receiver of the SGRes message keeps the SASucc message.
the node id of the sender. figure 3 shows an illustration of success in slot allotment
[Step 5:] If node i receives SGRes message from all of its where as figure 4 shows a case of failure in slot allotment.
114 JOURNAL OF COMMUNICATIONS SOFTWARE AND SYSTEMS, VOL. 15, NO. 2, JUNE 2019

According to figure 4, the node represented with black color

has not received SGRes message from two of its one-hop
neighbor nodes as the requested slot at these nodes are already 1 2 3 4 5 6 7 8 9
in the allotted(SA) state before receiving the SAReq message.
As a consequence, the node with black color updates the state
of that slot to un-allotted(SU). The nodes who have given 1
5 4 3 2 1
grant if did not receive the SASucc message within a certain
period of time then updates the state of the granted slot back
to un-alloted. 1 2 3 4 5 6 7

C. Phase3: Reallotment of Slots

Originally allotted slots Nodes re-allotted from a slot
In order to minimize the length of the feasible schedule, to another slot
Reduced slots after re-
the initially allotted slots(STA ) are reduced as per the ratio of allotment
Non-owner Nodes re-allotted
average two-hop neighbor count (NAvg ) to the maximum two- Finally allotted slots back to the original slot
hop neighbor count (NMax ). For example, If the ratio of NAvg
to the NMax is 1:2 then the number of reallotted slots to be
half of the currently alloted slots, i.e. STA . The process of slot Fig. 5. Slot reallotment for nodes
reallotment consists of following steps.
[Step 1:] Reduce the number of slots, i.e. SRA , based on
collision during data transmission, and the reduction in the
the ratio of average two-hop neighbor count (NAvg ) to the
length of the schedule (as compared to the feasible schedule)
maximum two-hop neighbor count (NMax ). Eventually, each
allows transmitting data with reduced latency.
node i checks whether it is eligible for reallotment or not by
comparing SAi with SRA . In case, SAi > SRA , the node i is
reallotted to a slot SAi = (Nmax -SAi )+1. IV. A NALYSIS OF THE HDSS A LGORITHM
[Step 2:] A node is referred as owner node to a slot, i.e. O(i,j) ,
where ith node alloted to j th slot, if it is initially alloted to We analyze our HDSS algorithm to show that the number
that slot. If a node is later reallotted to a slot then it is referred of slots allotted originally to prepare a feasible schedule can
as non-owner node for that slot, i.e. N(i,j) . be reduced and after such reduction in the number of allotted
[Step 3:] After the reallotment of slots, data transmission is slots still the schedule remain feasible.
carried out by each node i to verify which non-owner nodes, Let us consider a node Ni from the given figure 6. The cross
i.e. N(i,j) are in a collision during the same. In order to do marked nodes are considered as the one-hop neighbors of Ni .
so, each non-owner sensor node checks whether the medium The nodes with dots are considered as the two-hop neighbors
is free or busy using CSMA before their data transmission. of Ni . Let us assume that the number of two-hop neighbors of
Once the non-owner node finds the medium is busy then the Ni are Cnt2N . Initially, while preparing the feasible schedule,
collision will happen if the same node will transmit data packet the maximum of two-hop neighbors counts, i.e. Nmax is taken
along with the owner nodes. in to consideration. By considering the transmission range(
[Step 4:] The non-owner nodes, i.e. N(i,j) which are in a col- Tr ) and with basic assumption, i.e. the nodes are deployed
lision are reallotted back to their original slots. The reallotted in a random uniform manner, the Nmax can be computed as
slot information, i.e. RSi is broadcasted by the non-owner follows.
nodes to its one-hop neighbors. A sensor node that receives
the reallotted slot information updates the same at its own end N sq unit = N/Atot . (1)
if the received information is a new one, otherwise, it ignores N max = (π ∗ 2T r )2 ∗ N sq unit . (2)
the message. In case, the information is a new one, it again
broadcasts the updated information to its one-hop neighbors Initially, we assume that Nmax = Cnt2N as per the figure 6,
and so on. Finally, every node has the updated information of so there will be at least Nmax number of slots are required to
how many slots are currently available in the feasible schedule. prepare a feasible schedule, where STA = Nmax .
As per Figure 5, originally nine slots are available in the As per the figure 6, Let us consider two nodes named as
feasible schedule. After reallotment, the total number of slots Nj and Nk which are two-hop neighbors of node Ni and are
reduced to half, as the ratio of ”average two-hop density” to the assigned to the same slot in a feasible schedule. If both these
”maximum two-hop density” is 1:2. During data transmission, nodes start transmitting data packets at the same time still
out of the five allotted slots two of them i.e. slot number two these packets will not collide with each other even if they are
and four are in collision. Finally, the non owner nodes to whom the two-hop neighbors of each other as they are almost four-
the slot number two and four get allotted are now reallotted hop away from each other. Hence a feasible schedule can be
with slot number six and seven. Hence, the final number of prepared with STA = Nmax -1 number of slots.
slots allotted to handle collision become seven. It proves that there is a possibility of reducing the number
The whole process has been carried out once at the be- of originally allotted slots and at the same time it provides a
ginning to prepare the schedule. This schedule will handle feasible schedule.
M. R. LENKA et al.: ARDUINO-BASED SOLUTION FOR IN-CAR-ABANDONED INFANTS' CONTROLLING REMOTELY 115

Nj

Ni

NK

Node Ni
One hop-neighbors of Node Ni Fig. 8. With fixed node density, number of slots in collision after reallotment
Two hop-neighbors of Node Ni as compared to the total number of slots after reallotment

Fig. 6. Two-hop Neighbors of a node

Fig. 9. With fixed node density, time and energy spent for reallotment of
slots
Fig. 7. With fixed node density, average two-hop neighbors compared to
maximum two-hop neighbors
sensor nodes. With the increase in number of sensor nodes,
the number of nodes allotted to a particular slot increases.
V. S IMULATION E XPERIMENTS AND R ESULTS This leads to the possibility of the increase in collision due to
increase in number of sensor nodes.
The performance of our proposed algorithm is carried out Figure 9 shows the time and energy spent during slot
using castalia simulator. In this simulation, the sensor nodes reallotment. This result shows that the average time and energy
are uniformly distributed in a random manner. For this simula- spent almost remains same as the number of node increases
tion, the number of nodes deployed is varied from 50 to 1000 as the proposed algorithm is distributed in nature.
with a fixed as well as varying density. Various sensor node Figure 10 shows a comparison of the number of slots
parameters such as power transmission level, data transmission allotted to the final schedule in our proposed HDSS algorithm
rate,and energy consumption are taken into consideration as with the existing RD-TDMA and DRAND algorithm. Here,
per the information available in cc2420 data sheet and TelosB the result shows that the number of slots allotted to handle
data sheet. the collision in our proposed HDSS algorithm is less as
Figure 7 shows the computed average two-hop neighbors compared to the RD-TDMA and DRAND algorithm, i.e. our
vs the actual two-hop neighbors in different size network. proposed HDSS algorithm performs better than RD-TDMA
In this figure, the average two-hop neighbors remain almost and DRAND algorithm w.r.to the number of slots allotted to
50% of the actual two-hop neighbors. As the average two-hop handle collision.
neighbors are almost half of the actual two-hop neighbors, Figure 11 shows a comparison of the number of allotted
hence reduction of the originally allocated slots to it’s half slots, reallotted slots, and number of slots in collision after
will improve the performance. Our proposed algorithm proves reallotment. It shows that the number of slots which are in
the same. collision almost remains the same with varying area. As the
Figure 8 shows the number of slots are in collision after number of nodes remains same hence, the chance of collision
reallotment. From this figure, it shows that the number of also remains the same with varying number of nodes.
slots in collision increases with the increase in number of Figure 12 shows that the latency in data transmission is
116 JOURNAL OF COMMUNICATIONS SOFTWARE AND SYSTEMS, VOL. 15, NO. 2, JUNE 2019

Fig. 10. With fixed node density, number of slots allotted for the feasible Fig. 12. Latency in data transmission in proposed HDSS, DRAND, and RD-
schedule using proposed HDSS, DRAND, and RD-TDMA algorithm TDMA algorithm

Fig. 11. With varying node density, comparison of the number of allotted Fig. 13. Convergence time for slot reallotment
slots, reallotted slots, and slots in collision after reallotment

less in our proposed HDSS algorithm as compared to the RD- VI. C ONCLUSION
TDMA and DRAND. In case of the HDSS algorithm as less
number of slots are used to prepare the feasible schedule as
compared to RD-TDMA and DRAND hence the result. In this paper, we have proposed a slot scheduling algorithm,
named HDSS, for wireless sensor networks to handle collision
during communication. In the first phase of the proposed
TABLE V
AVERAGE E NERGY SPENT ( IN JOULES ) FOR VARIOUS P HASES DURING
algorithm, two-hop neighbors are calculated for each node in
P REPARATION OF THE F EASIBLE S CHEDULE WSN. Based on the maximum two-hop neighbors count over
the whole network, the total number of slots for a feasible
Phases HDSS DRAND
Neighbor Discovery 1.01 0.93
schedule is decided. In the next phase, reallotment of slots
Slot Allotment 5.12 5.15 is done by reallocating the nodes from the last to first , last
Slot Re-allotment 0.02 - but one to the second and so on. Finally, after reallotment,
the number of slots in collision is found out and accordingly
the final schedule is prepared. This reduces the number of
Table V shows that the proposed HDSS algorithm con- slots allocated for handling collision. The efficiency of the
sumes almost the same amount of energy as compared to the proposed algorithm has been evaluated with fixed as well as
DRAND. Most importantly, with the expense of same amount varying node density. Our proposed algorithm is compared
of energy, the proposed algorithm can able to reduce the with an existing distributed slot scheduling algorithm called
number of slots to prepare the feasible schedule as compared RD-TDMA. Comparison results show that the proposed HDSS
to DRAND. algorithm outperforms the existing RD-TDMA algorithm to
Figure 13 shows that with increase in number of sensor handle the collision with reduced number of slots and at the
nodes the convergence time for reallotment of slots increase same time improves the efficiency of data communication in
very slowly. the channel with reduced latency.
M. R. LENKA et al.: ARDUINO-BASED SOLUTION FOR IN-CAR-ABANDONED INFANTS' CONTROLLING REMOTELY 117

R EFERENCES [20] Fafoutis X, Mauro A, Vithanage M, Dragoni N. Receiver-initiated

medium access control protocols for wireless sensor networks. COMM
[1] Ye W, Heidemann J, Estrin D. An energy-efficient MAC protocol for COM INF SC 2015; 76: 55-74, DOI: 10.1016/j.comnet.2014.11.002.
wireless sensor networks. In: Proceedings of the IEEE Infocom; 23-27 [21] Segun A, Ao A, Eo O. A survey of medium access control protocols in
June 2002; New York, NY, USA. New York, NY, USA: IEEE. pp. 1567- wireless sensor network. International Journal of Computer Applications
1576, DOI:10.1109/INFCOM.2002.1019408. 2015; 116: 1-8, DOI: 10.5120/20465-9980.
[2] Dam T, Langendoen K. An adaptive energy-efficient MAC protocol [22] Rhee I, Warrier A, Aia M, Min J, Xu L. DRAND: Distributed random-
for wireless sensor networks. In: Proceedings of the 1st International ized TDMA scheduling for wireless ad hoc networks. IEEE T MOBILE
Conference on Embedded Networked Sensor Systems; 05-07 November COMPUT 2009; 8: 1384-1396, DOI: 10.1109/TMC.2009.59.
2003; Los Angeles, California, USA. New York, NY, USA: ACM. pp. [23] Slama I, Shrestha B, Jouaber B, Zeghlache D, Erke T. DNIB: Distributed
171-180, DOI:10.1145/958491.958512. neighborhood information based TDMA scheduling for wireless sensor
[3] Li S, Qian D, Liu Y, Tong J. Adaptive distributed randomized TDMA networks. In: Proceedings of 2008 IEEE 68th Vehicular Technology
scheduling for clustered wireless sensor networks. In: Proceedings of Conference; 21-24 September 2008; Calgary, BC, Canada. New York,
2007 International Conference on Wireless Communications, Networking NY, USA: IEEE, DOI: 10.1109/VETECF.2008.25.
and Mobile Computing; 21-25 September 2007; Shanghai, China. New [24] Ahmad A, Hanzálek Z. Distributed Real Time TDMA Scheduling
York, NY, USA: IEEE. pp. 2688-2691, DOI:10.1109/WICOM.2007.668. Algorithm for Tree Topology WSNs. IFAC-PapersOnLine 2017. DOI:
[4] Bhatia A, Hansdah R. RD-TDMA: A randomized distributed TDMA 10.1016/j.ifacol.2017.08.1484.
scheduling for correlated contention in WSNs. In: Proceedings of 2014 [25] Lee J, Cho S. Tree TDMA MAC Algorithm Using Time and Frequency
28th International Conference on Advanced Information Networking and Slot Allocations in Tree-Based WSNs. Wireless Personal Communica-
Applications Workshops; 13-16 May 2014; Victoria, BC, Canada. New tions 2017. DOI: 10.1007/s11277-016-3938-9.
York, NY, USA: IEEE, DOI: 10.1109/WAINA.2013.23. [26] Danmanee T, Nakorn K, Rojviboonchai K. CU-MAC: A Duty-Cycle
[5] Lenka M, Swain A, Sahoo M. Distributed slot scheduling algorithm for MAC Protocol for Internet of Things in Wireless Sensor Networks. ECTI
hybrid CSMA/TDMA MAC in wireless sensor networks. In: Proceedings Transactions on Electrical Engineering, Electronics, and Communications
of 2016 IEEE International Conference on Networking, Architecture and 2018.
Storage (NAS); 8-10 August 2016; Long Beach, CA, USA. New York,
NY, USA: IEEE, DOI:10.1109/NAS.2016.7549396.
[6] Ergen S, Varaiya P. TDMA scheduling algorithms for wireless sensor
networks. WIREL NETW 2010; 16: 985–997, DOI:10.1007/s11276-009- Manas Ranjan Lenka received his M.Tech de-
0183-0. gree in Computer Science and Engineering from
[7] Dobslaw F, Zhang T, Gidlund M. End-to-End reliability-aware scheduling Biju Patnaik University of Technology (BPUT),
for wireless sensor networks. IEEE T IND INFORM 2016; 12: 758-767, Bhubaneswar, India in 2010. Currently, he is work-
DOI: 10.1109/TII.2014.2382335. ing as an assistant professor in the School of Com-
[8] Mao J, Wu Z, Wu X. A TDMA scheduling scheme for many-to- puter Engineering, KIIT Deemed To Be University,
one communications in wireless sensor networks. COMPUT COMMUN Bhubaneswar, India. His research interests include
2007; 30: 863-872, DOI: 10.1016/j.comcom.2006.10.006. wireless sensor networks, Internet of Things.
[9] Yang D, Xu Y, Wang H, Zheng T, Zhang H, Zhang H, Gidlund M.
Assignment of segmented slots enabling reliable real-time transmission
in industrial wireless sensor networks. IEEE T IND ELECTRON 2015;
62: 3966-3977, DOI: 10.1109/TIE.2015.2402642.
[10] Guo L, Li Y, Cai Z. Minimum-latency aggregation scheduling in Amulya Ratna Swain received his M.E. Degree
wireless sensor network. J COMB OPTIM 2016; 31: 279–310, DOI: in Software Engineering from Jadavpur University,
10.1007/s10878-014-9748-7. Calcutta, India in 2006. He received the PhD de-
[11] Ramanathan S, Lloyd E. Scheduling algorithms for multihop ra- gree in Computer Science from Indian Institute of
dio networks. IEEE ACM T NETWORK 1993; 1: 166-177, DOI: Science, Bangalore, India, in 2013. Currently, he
10.1109/90.222924. is working as an associate professor in the School
[12] Wang Y, Henning I. A deterministic distributed TDMA scheduling of Computer Engineering, KIIT Deemed To Be
algorithm for wireless sensor networks. In: Proceedings of 2007 Interna- University, Bhubaneswar, India. His research inter-
tional Conference on Wireless Communications, Networking and Mobile ests include wireless sensor networks, distributed
Computing; 21-25 September 2007; Shanghai, China. New York, NY, computing and operating systems.
USA: IEEE. pp. 2759-2762, DOI: 10.1109/WICOM.2007.685.
[13] Gandham S, Zhang Y, Huang Q. Distributed time-optimal scheduling for
convergecast in wireless sensor networks. COMM COM INF SC 2008;
52: 610-629, DOI: 10.1016/j.comnet.2007.10.011. Biraja Prasad Nayak received his M. Tech. degree
[14] Rhee I, Warrier A, Aia M, Min J, Sichitiu M. Z-MAC: a hybrid MAC for in Computer Science and Engineering from School
wireless sensor networks. IEEE ACM T NETWORK 2008; 16: 511-524, of Computer Engineering, Kalinga Institute of Indus-
DOI: 10.1109/TNET.2007.900704. trial Technology (KIIT) Deemed To Be University,
[15] Slama I, Jouaber B, Zeghlache D. Priority-based hybrid MAC for energy Bhubaneswar, Odisha, India in 2017. His research
efficiency in wireless sensor networks. Wireless Sensor Network 2010; interests are Wireless Sensor Networks and Mobile
2: 755-767, DOI: 10.4236/wsn.2010.210091 . Computing.
[16] Zhuo S, Song Y, Wang Z, Wang Z. Queue-MAC: A queue-length aware
hybrid CSMA/TDMA MAC protocol for providing dynamic adaptation to
traffic and duty-cycle variation in wireless sensor networks. In: Proceed-
ings of 2012 9th IEEE International Workshop on Factory Communication
Systems; 21-24 May 2012; Lemgo, Germany. New York, NY, USA: IEEE,
DOI: 10.1109/WFCS.2012.6242552.
[17] Ergen S, Varaiya P. PEDAMACS: Power efficient and delay aware
medium access protocol for sensor networks. IEEE T MOBILE COMPUT
2006; 5: 920–930, DOI: 10.1109/TMC.2006.100.
[18] Sun Y, Du S, Gurewitz O, Johnson D. Dw-mac: a low latency, energy
efficient demand-wakeup mac protocol for wireless sensor networks.
In: Proceedings of the 9th ACM international; 26-30 May 2008; Hong
Kong, Hong Kong, China. New York, NY, USA: ACM. pp. 53-62, DOI;
10.1145/1374618.1374627.
[19] Oller J, Demirkol I, Casademont J, Paradells J, Gamm G, Reindl L.
Has time come to switch from duty-cycled MAC protocols to wake-up
radio for wireless sensor networks?. IEEE ACM T NETWORK 2016; 24:
674-687, DOI: 10.1109/TNET.2014.2387314.