Sage Journals: Discover world-class research

Abstract

IEEE 802.15.4 standard is widely used as a communication protocol by low-powered devices, including Internet of Medical Things–enabled technologies. These low-powered technologies have the ability to integrate several communicating and interacting devices for the purpose of wide range applications such as Smart homes and healthcare applications. However, the fixed superframe structure of IEEE 802.15.4 results in the unequal resource utilization among these low-powered heterogeneous Internet of Things devices. The low resource utilization degrades the network performance. In order to improve the network performance and optimum resource utilization, dynamic superframe approach is recommended. This article presents a content-based dynamic superframe adaptation algorithm for the low-powered Internet of Medical Things devices to address the resource utilization challenges. In the content-based dynamic superframe adaptation, the network coordinator dynamically adjusts the superframe along with the backoff exponent. It uses five Quality of Service metrics simultaneously: the application defined data traffic, receive ratio, Personal Area Network source nodes in numbers, number of collisions, and observed delay to achieve the optimal solution. A detailed analysis of the simulations shows that the content-based dynamic superframe adaptation is able to behave more intelligently to adjust superframe dynamic allocations according to the application’s content requirements. Moreover, it outperforms the other existing schemes in achieving the better resource utilization in IEEE 802.15.4 framework.

Keywords

IEEE 802.15.4 low-rate wireless personal area network Internet of Medical Things superframe adaptation duty cycle traffic heterogeneity healthcare applications backoff adjustment beacon order and superframe order

Introduction

Recently, the new and advanced technologies have changed the traditional way of 39 operations and almost every field of life changed into new integrated and smart technologies such as smart transportation systems,^1,2 smart health systems,³ agriculture precision,^4,5 Internet of Things (IoT),^6,7 and Internet of Vehicles (IoV).⁸ The most important premises of IoT is to connect devices and sensors without human intervention. According to a new data from the juniper research,¹ the number of IoT connected devices in future will have an increase of more than 28.5% from 13.4 billion at the end of 2015 and 38.5 billion at the end of 2020. The IoT is getting prominence in the identification and sensing technology due to its small size, low power, low price, light weight, and inexpensive maintenance rates. It is going to be used in many advanced fields like smart homes, medical, education, pharmaceuticals, manufacturing, and retail as shown in Figure 1. The major attributes of Internet of Medical Things (IoMT) network are the heterogeneity of network technologies, heterogeneity of communicating devices, and the interoperability among these heterogeneous devices. Researchers from academia and industry have realized that the deterring of major developments in IoT is the lack of standards for the IoT architecture. Therefore, IEEE 802.15.4 is extensively used as the premier choice medium access protocol for IoT applications due to IPv6 addressing. The focus of this article is also on the use of IEEE 802.15.4 with the aim to improve the network performance and life time by efficient handling and distribution of resource utilization. It is very difficult to replace the communicating devices, particularly the wearable health monitoring sensors in IoT paradigm.⁹

Figure 1.

IoMT infrastructure.

According to standard IEEE 802.15.4, a network can be formed by two types of devices: fully functional device (FFD) and reduced functional device (RFD). FFDs hold all the necessary operation of IEEE 802.1.5.4 stack for establishing and maintaining a network.¹⁰ Moreover, it has more computational capability and energy resources to become a Personal Area Network (PAN) coordinator as compared to RFD devices. RFD supports limited operations and cannot be used to initiate and manage a network; however, it can be used to perform simple sensing tasks. A wireless personal area network (WPAN) network can be arranged in a systematic way to form star or cluster tree. Cluster tree connectivity is useful to create extended and large networks such as mesh topologies. Star topology is a suitable choice for short-range small networks such as agriculture¹¹ and industrial automation.¹² Star topology consists of at least one FFD referred to as PAN coordinator and connecting devices (FFDs and RFDs). The PAN coordinator is responsible for the dispatch of superframe structure. A superframe structure is a time referencing used by FFD for communicating with RFDs. It further consists of beacons, data, and acknowledgment time phases. When a device receives the beacons, they become functional and start to deliver the contents within the superframe boundary. Superframe structure identifies the active period inside which source nodes can send data and can be active till the arrival of the next superframe. On the completion of active period, source nodes will remain in the inactive portion as specified in superframe duration till next beacon interval (BI). PAN coordinator frequently transmits beacons to its connected nodes for the purpose of synchronization and association and to guarantee that nodes match with the same source. Superframe structure plays a vital role for the duty cycle of nodes. However, the current standard recommends the use of fixed superframe structure, which is manually tuned by the network administrator during network deployment phase. However, it does not fulfill every application demands. In order to make it suitable for most of the applications, several approaches recommend to dynamically schedule the IEEE 802.15.4 superframe structure in beacon-enabled mode based on some traffic approximations like end-to-end latency, collision ratio (CR), occupancy ratio, number of packets received (PR) at PAN coordinator or traffic queue.

In this article, we propose a content-based dynamic superframe adaptation (CDSA) scheme to adapt both beacon order (BO) and superframe order (SO) simultaneously based on the content requirements. The content requirements are application specific, consisting of two major entities: data rate for successful detection of an event and the delay bound within which the information should reach the PAN coordinator. The proposed superframe adjustment scheme focuses on providing higher network throughput and reducing end-to-end latency during event reporting. Moreover, the proposed superframe adaptation scheme aims to enhance the network lifetime by reducing the duty cycle of nodes during no network activity.

The rest of the article is organized as follows: “Related work” section presents the extensive study of related dynamic superframe adjustment. However, CDSA algorithm network model is presented in “System model” section. “Proposed methodology” section outlines the working of CDSA algorithm. A detailed analysis and discussion are described in “Performance evaluation” section, whereas in the last section conclusion and future research have been presented.

Related work

There are some existing superframe adaptation schemes. An Adaptive Algorithm to Optimize the Dynamics (AAOD)¹³ aims to alter the superframe duration (SD) to maximize the packet delivery ratio. The central controller (PAN coordinator) calculates the number of packets and then compares with previously used SO in the network. If PAN coordinator observes higher packets than a pre-defined threshold, the algorithm increases the SO. AAOD¹³ is silent on the use of other important parameters like traffic loads and collision. An Adaptive MAC protocol (AMPE)¹⁴ adjusts the duty cycle of the nodes based on the occupancy ratio of SD. Initially, the PAN coordinator computes superframe occupancy ratio (OR) and associates the same with two fixed application-specific threshold values which are different from each other. These fixed threshold values are upper and lower superframe occupation ratio. In case the computed occupancy ratio of superframe is larger than the upper limit of the threshold, then the PAN coordinator increments SO by 1. In case of less than the lower threshold, the coordinator adjusts SO by decreasing it.

Dynamic BO and SO Adaptation Algorithm (DBSAA)¹⁵ schedules the duty cycle dynamically of both BO and SO simultaneously. PAN coordinator adapts both BO and SO. It is based on the four estimation parameters: superframe OR, collision rate (CR), PR by the PAN coordinator, and number of transmitting end devices. In DBSAA,⁶ the network load is calculated first, and then it computes the changes occurred during network activity, and finally after $α$ (decision parameter) intervals, the DBSAA algorithm is triggered. DBSAA shows high network performance, as it adjusts both BO and SO simultaneously. Nevertheless, DBSAA considers the same type of data and does not take into account the time bound or emergency data during event reporting. A modified superframe structure in Sabita et al.¹⁶ is proposed for periodic and emergency traffic for Wireless Body Area Network (WBAN) based on the star topology. The objectives of this protocol are to enhance the throughput and lower the latency. The proposed MAC protocol¹⁶ modified the frame format by introducing emergency contention period (ECP), advertisement beacon (AB), periodic contention period (PCAP), notification beacon (NB), and data transmission period (DTP). In ECP portion, if any node has emergency data to send, the node sends an emergency alarm to the coordinator. As a result, the coordinator broadcasts a “set” flag using AB which indicates the presence of emergency traffic; otherwise, the flag is set to “reset” in case of no emergency data. On the contrary, if the end node receives a “set” flag, a DTS is sent to the coordinator in order to transmit regular traffic just after emergency traffic. The proposed protocol shows enhanced network throughput and lower delay; however, definitely, the energy consumption of the node increases due to frequent listening in the idle period and data transmission. Adaptive Energy Efficiency Algorithm (ABSD)⁸ aims to optimize the energy and Quality of Service (QoS) for packet transmission. Primarily, the proposed algorithm estimates the traffic load by computing the number of incoming packets $T [cur]$ in $β$ interval, where $β$ is the number of wait interval before triggering the algorithm. Then, the PAN coordinator compares $T [cur]$ with the T[pre] values observed in the previous $β$ BIs. After these analysis, the PAN coordinator triggers the ABSD algorithm to schedule the BO and SO. Furthermore, the adjusted values are examined to improve the correctness of both BO and SO values. The PAN coordinator collects information of the initial queue length $Qi$ and the occupied queue $Q_{(ioc)}$ for all $i \in S$ where S refers to the number of sensor nodes. This information allows the PAN coordinator to accurately determine the remaining queue, Q_(ioc) = Qi – Q_(ioc). Here, they define a term of queue Statex, Q(iStatex) = Q_(ioc)/L, which can be used to evaluate whether a node can buffer more packets’ combination between $Q (iStatex)$ and incoming traffic $1 / λ_{i}$ , where $λ i$ is the packet arrival time at each node, and then the coordinator adjusts the values of both BO and SO for the next superframe.

Another modified superframe structure presented in Dahae et al.¹⁷ aims to improve the energy efficiency by reducing the clear channel assessment (CCA) and by increasing the sleep duration. The authors introduced four new phases in the existing superframe structure: notice, scheduling, reserved, and non-reserved. In notice phase, the node notifies the coordinator regarding its data size before sending. In scheduling phase, the coordinator sets the size for data transmission. On receiving the allocation time, the node goes into sleep mode. Reserved phase is used by the node to receive its transmittable time; on the completion of data transmission, the end device goes into sleep mode. On the contrary, during non-reserved phase, all devices stay awake all the time to transfer the data. The SD adjustment scheme (SUDAS)¹⁸ adjusts the length of the guaranteed time slot (GTS) based on the interval of the packet size. This work considers a star topology that has seven end devices connected to a single PAN with BO and SO set to equal length. However, by allocating GTS, the normal traffic can be sacrificed and the delay of communication will be increased. This article extends our previous work titled “A delay mitigation dynamic scheduling algorithm (DMDSA)” presented in Sorell.¹⁹

System model

In this section, basic network-related assumptions, definitions, and network model used by CDSA are explained. This work is based on a star topology–based IoT architecture for healthcare applications. The adjustment of the controlled superframe parameters BO and SO is referred to as superframe adaptation. The work is that such adjustment is responsible for the selection of BI and active portion within a BI. The use of BO and SO in an efficient manner improves the resource utilization of low-power devices. It gently prioritizes the traffic and introduces the wake–sleep phenomenon. The proposed algorithm is developed to consider the content-based approach for dynamic superframe adjustment of low-power IoT healthcare. However, the content requirements such as data rate and delay deadlines are not the same in different applications.

In this article, an IoT architecture is focused in which the WPANs for healthcare having various in- and on-body sensors installed to monitor events such as blood pressure, temperature, electrocardiogram (ECG) and implantable cardiac defibrillator (ICD). In all these applications, the nature of data is dissimilar and reporting rate of each event is different from the other. In addition, WPAN deployed for industrial and home automation has periodic data and requires low data rates. The coordinators in the existing duty cycle adjustment schemes^13–20 predict data rate for the next superframe interval based on observed data rates in the current interval. The coordinator is unaware of the amount of traffic that could be generated by an event. Therefore, it cannot precisely adjust the superframe. Also, the end-to-end delay requirement of different contents is not known to the coordinator. In the proposed scheme, it is assumed that the PAN coordinator knows about the content’s nature that is delivered by low-power sensors. Moreover, the content requirement in terms of delay deadline is also known to the coordinator. This kind of information can be exchanged between a PAN member and a PAN coordinator during the association phase in which a device becomes a PAN member.

In CDSA, the dynamic adjustment of superframe after every BI is the responsibility of coordinator. At network initialization and during idle operations of network, the default values of BO and SO are essential to define. To decrease the wastage of energy during idle network operations, it is recommended that smaller value of SO should be smaller than BO. During periods of data reporting, CDSA calculates the size of forthcoming active duration based on the observed traffic within the current interval. Hence, it measures the expected SO $(S O_{(Exp)})$ for every BI after the expiry. $S O_{(Exp)}$ is the expected length of SD for the next BI and is calculated for individual traffic flows having same content requirement. $S O_{(Exp)}$ depends on number of sources, data rates of individual source nodes, and the time required to transmit a single packet. $S O_{(Exp)}$ is estimated by equation (1)

$\begin{matrix} S O_{(Exp)} = (PA N_{(src)} \times NPk t_{(SD)} \times Pk t_{(size)} \times Pro p_{t}) \\ \times Tran s_{(delay)} \end{matrix}$ (1)

$Pro p_{t} = \frac{Pk t_{(size)}}{250}$ (2)

$Tran s_{(delay)} = T_{(ACK)} + ACK + T_{(IFS)} + (2 CCA)$ (3)

where $PA N_{(src)}$ is the total number of source nodes transmitting at the same event (with the same packet size) in an interval, $Pk t_{(size)}$ is the size of packets in bits, $NPk t_{(SD)}$ is the number of packets that a single-source node will generate during the SD, and $Pro p_{t}$ is the propagation time for a single packet. Here maximum link rate is 250 Kbps. $Tran s_{(delay)}$ incorporates the time required for medium access using CCA considering no contention, inter-frame spacing (IFS), and the time to send ACK. Apart from $S O_{(Exp)}$ , different other parameters are used in CDSA for dynamic adjustment, including packet receive ratio (RR), packet CR, and observed end-to-end delay ( $dela y_{(obs)}$ ). The rest of this section explains these terms in detail. RR is computed using individual traffic flows of similar requirements and is calculated using equation (4)

$RR = \frac{Tota l_{(pkt)}}{PA N_{(src)} \times NPk t_{(SD)}}$ (4)

where $Tota l_{(pkt)}$ is the total number of received packets at PAN coordinator; CR is the amount of packets dropped to the amount of packet transmitted during the active SD and is calculated using equation (5)

$CR = \frac{(PA N_{(src)} \times NPk t_{(SD)}) - (Tota l_{(pkt)})}{(PA N_{(src)} \times NPk t_{(SD)})}$ (5)

The average $dela y_{obs}$ for all packets having same delay bound received at the PAN coordinator during BI is calculated using equation (6). The average end-to-end delay includes queuing delay, medium access delay, and propagation delay

$Dela y_{obs} = \frac{\sum (T_{recv} - T_{sent})}{NPk t_{SD}}$ (6)

where $T_{recv}$ is the time coordinator receives a packet and $T_{sent}$ is the time when the source node’s MAC layer receives the packet from upper layer.

Proposed methodology

The operation of the proposed CDSA scheme is based on the calculation of $S O_{Exp}$ and the value of $S O_{Curr}$ observed during the current BI. $S O_{Exp}$ is computed at the last of each BI and it predicts the required value of SO for the next BI. It is based on the number of transmitting low-power IoT devices and their data rates. The difference in the expected and the current value of active duration is adjusted using the proposed CDSA algorithm. Ideally $S O_{Exp}$ is the duration that should be used for the next interval; however, it is important to observe that $S O_{Exp}$ meets the content requirements. The network performance is monitored in terms of the packet received ratio and communication delay observed at the coordinator. If the required network performance is not achieved by using current $S O_{Exp}$ , it would be adjusted for the next BI. The major reason $S O_{Exp}$ does not satisfy the content requirement every time is that the traffic generated is bursty in nature and it suddenly increases the contention. This results in packet drops and increases the end-to-end delay. Therefore, in certain scenarios, $S O_{Exp}$ is further increased to mitigate the contention. Moreover, other parameters such as BO and backoff exponent (BE) are adjusted to meet the network’s traffic flow requirements.

The symbols used in CDSA algorithm are listed in Table 1. The algorithm is triggered at the coordinator upon the expiry of every BI. There are two basic possibilities: $S O_{Exp}$ is either greater than $S O_{Cur}$ or less than the observed $S O_{Cur}$ .

Table 1.

Symbols used in CDSA algorithm and their description.

Symbols	Description
$S O_{Exp}$	Expected SO for the next BI
$S O_{Cur}$	SO value used in the CI
$S O_{Default}$	Default value of the SO
$B O_{Default}$	Default value of the BO
$B E_{Cur}$	Current value of the BE
$B E_{Default}$	Default value of the BE
N	Number of CI, incremented after every BI
A	Indicates no. of wait intervals after which superframe adjustment decision revisited
RR	Packet RR at coordinator during the BI
$Dela y_{obs}$	Observed end-to-end delay in the last BI
$Dela y_{Req}$	Application defined deadline associated with delivery

CDSA: content-based dynamic superframe adaptation; SO: superframe order; BI: beacon interval; CI: current interval; BO: beacon order; BE: backoff exponent; RR: receive ratio.

Condition I: SO_Exp > SO_Cur

When $S O_{Exp}$ is greater than $S O_{Cur}$ , SO is immediately readjusted to $S O_{Exp}$ for the next interval. A greater value of $S O_{Exp}$ shows that the $S O_{Cur}$ is not properly adjusted to accommodate the traffic generated by source nodes in the upcoming interval. The computation of $S O_{Exp}$ is independent of BO; therefore, it is possible that the current value of BO can be less than $S O_{Exp}$ . In this case, BO is recomputed to adjust the SD duration. Moreover, it is ensured that BO does not go beyond the default upper limit defined by IEEE 802.15.4 or the limit defined as per application requirements during all scenarios. If the $S O_{Cur}$ is lower than BO after adjustment, then BO is decreased to reduce the latency. The reason is that BO can be at default value at this instant which is large in number as the applications can select large default values for minimizing the energy consumption during idle network time.

Condition II: SO_Exp≤ SO_Cur

The $S O_{Cur}$ is readjusted to the $S O_{Exp}$ for decreasing the duty cycle of network nodes ideally when $S O_{Exp}$ is less than or equal to the $S O_{Cur}$ . However, before adjusting the $S O_{Cur}$ , it is necessary to observe the content requirements of the network that are measured in terms of required packet delivery ratio and tolerable delay. If the content requirements are not fulfilled by the $S O_{Cur}$ , then decreasing the SO will result in further degradation of network performance. $S O_{Exp}$ is a theoretically calculated value that is unable to immediately adjust to actual network requirements in case of sudden traffic flow within the network. Traffic in event monitoring networks such as WPAN is bursty in nature and can suddenly increases the contention time for medium access within the PAN members. This results in large packet drops and high latency of communication. In the aforementioned cases, meeting deadlines of reported events is especially difficult by the $S O_{Exp}$ . The CDSA algorithm in given algorithm 1 strives to fulfill the application-specific requirements in network optimization phase (NOP). The CDSA is in NOP when $S O_{Exp}$ is equal to $S O_{Cur}$ . During the NOP, the content requirements are checked first and then $S O_{Cur}$ , BO, or BE is adjusted accordingly. Instantly, few BIs are essential for the current network settings to improve the network performance. Therefore, the CDSA enters into NOP, if the network persists in NOP for more than application-defined ( $α$ ) consecutive BI. During the NOP phase, the expected receive ratio (RR_Exp) of the content at the PAN coordinator is observed which can be below the acceptable application threshold. It is because of sudden increase in the contention, interference, and congestion due to the introduction of new traffic flows within the previous BI. $S O_{Cur}$ is aggressively increased when both PR ratio and delay requirements are not satisfied. However, if the delay deadlines are met or the traffic has no deadline requirements, $S O_{Cur}$ is increased slowly. In the NOP phase, it is possible that RR_Exp is above the required application’s defined level. To optimize the network performance, CDSA further checks the delay requirements. SO is further increased with the decrease in BE for delay bound traffic when the delay requirements are not satisfied. BE is slowly increased to its default value when delay requirements are also met.

Algorithm 1. CDSA Algorithm1.
procedure At Each Beacon Interval
Require: SO and BO
if $S O_{Exp}$ > $S O_{Cur}$ then
$S O_{Cur} \leftarrow S O_{Exp}$
if $S O_{Cur}$ > $BO$ then
$BO \leftarrow S O_{Cur}$
else
if BO − $S O_{Cur}$ >= 1 then
$BO \leftarrow BO$ - $1$
end if
end if
else
if $S O_{Cur}$ <= $S O_{Default}$ then
BO< $B O_{Default}$
$BO \leftarrow B O_{Default}$
$S O_{Cur} \leftarrow S O_{Default}$
else
if BO − $B O_{Default}$ >= 1 then
$BO \leftarrow BO$ - $1$
$S O_{Cur} \leftarrow S O_{Cur}$ + 1
end if
end if
else
if $RR$ % $α$ = 0 then
if $RR$ >= $Recv_TH$ then
if $Dela y_{Obs}$ >= $Dela y_{Req}$ then
$S O_{Cur} \leftarrow S O_{Cur}$ + 1
$B E_{Cur} \leftarrow B E_{Cur}$ − 1
else
$S O_{Cur} \leftarrow S O_{Cur}$ − 1
$B E_{Cur} \leftarrow B E_{Default}$
end if
else
if $Dela y_{Obs}$ >= $Dela y_{Req}$ then
$S O_{Cur} \leftarrow S O_{Cur}$ + 2
$B E_{Cur} \leftarrow B E_{Cur}$ + 1
else
$S O_{Cur} \leftarrow S O_{Cur}$ + 1
$B E_{Cur} \leftarrow B E_{Cur}$ + 1
end if
end if
end if
end if

Performance evaluation

This section discusses the detailed performance evaluation of CDSA in comparison with AAOD¹³ (adjusts SO only), AMPE¹⁴ (adjusts SO only), and DBSAA¹⁵ (adjusts both BO and SO). The simulation analysis is performed in star topology–based IEEE 802.15.4 beacon-enabled network using NS-2.¹¹ The PAN coordinator is placed at the middle of the network range and the nodes are randomly distributed in a 15-m coverage area. These low-power IoT devices are deployed with the purpose of healthcare application requirements. The resource utilization based on the healthcare application was the main focus of the proposed CDSA. In CDSA algorithm, $α$ is set to 3. Therefore, we want to keep the past history of the last three BI is kept as already to the simulation. This duration is adequate to notice a noteworthy change in the network load. For realistic results, the energy model of nodes is mirrored according to ATMEL mote.¹² The remaining simulation details are given in Table 2. Average results are obtained by running multiple simulations. Subsequently, evaluation results for throughput, packet delivery ratio, latency, and energy consumption are also calculated and presented in the forthcoming section.

Table 2.

Simulation parameters.

Parameters	Value
Time of simulation	100 s
Radio range	15 m
Frequency	2.45 GHz
$B O_{Default}$	6 (CDSA parameters)
$B O_{Maximum}$	10 (CDSA parameters)
$S O_{Default}$	2 (CDSA parameters)
macBeaconOrder	6 (IEEE 802.15.4, AMPE, and DBSAA)
macSuperframeOrder	2 (IEEE 802.15.4, AMPE, AAOD,and DBSAA)
Traffic type	CBR
Interface queue length	30
Packet size (B)	60
No. of PAN members	1–20
Data rate (Kbps)	100–250
Primary energy	4 J
Energy model	Atmel Mote

CDSA: content-based dynamic superframe adaptation; AMPE: adaptive MAC protocol; DBSAA: Dynamic Beacon order and Superframe order Adaptation Algorithm; PAN: Personal Area Network.

Packet delivery ratio

The Packet delivery ratio of CDSA and IEEE 802.15.4 at data rate 100 and 250 Kbps is presented in Figure 2. Here, CDSA uses BO = 6, SO = 2, while IEEE 802.15.4 uses BO = 6, SO = 2, 3, 4, and 5. It is observed that CDSA packet delivery ratio is more than the default IEEE 802.15.4; in each case, CDSA dynamically adjusts SO as per application requirements. On the contrary, for IEEE 802.15.4, the packet delivery ratio increases when the SO increases. It is evident that at SO = 5 and BO = 6, the duty cycle of nodes is high and the CAP duration is more enough. It results in an increase in the performance of IEEE 802.15.4. The packet delivery ratio for CDSA, IEEE 802.15.4, AAOD,¹³ AMPE,¹⁴ and DBSAA¹⁵ at the maximum data rate 250 Kbps, BO = 6 and SO = 2 is shown in Figure 3. In this scenario, 10 source nodes generate traffic at 25 Kbps per node. The default active portion of CAP (SO = 2) in IEEE 802.15.4 is too small to support traffic flows and results in low delivery rate. CDSA provides higher delivery ratio as compared to other strategies. It not only adjusts BO and SO but amends the BE according to network traffic.

Figure 2.

Packet delivery ratio of CDSA and IEEE 802.15.4 at 100 and 250 Kbps using BO = 6 and SO = 2.

Figure 3.

Packet delivery ratio of CDSA and other algorithms at 250 Kbps using BO = 6 and SO = 2.

Network throughput

Figure 4 compares the average throughput of CDSA, IEEE 802.15.4 LR-WPAN, AAOD,¹³ AMPE,¹⁴ and DBSAA.¹⁵

Figure 4.

Throughput of CDSA and other protocols against simulation time.

Five source nodes periodically transmit data for 100 s only. It then stops data reporting for 100 s to the PAN coordinator where the data rate is 20 Kbps per source node. Initially, the network load is gradually increased as nodes report data to the PAN Coordinator. The CDSA performs better than the other protocols because CDSA adapts both BO and SO simultaneously based on the content requirements.

To support the claim that CDSA is dynamically adjusting both BO and SO, Figure 6 shows the observed BO and SO values using CDSA. The default value of BO and SO is 6 and 2, respectively. The maximum threshold set for BO was 10, whereas for SO, it is set to a value of 2. As shown in Figure 5, CDSA switches BO and SO to default when network is idle, while during data reporting period, they are adjusted according to traffic requirements.

Figure 5.

Dynamic BO and SO behavior of CDSA algorithm.

Figure 6 indicates the average throughput of CDSA, IEEE 802.15.4, AAOD,¹³ AMPE,¹⁴ and DBSAA¹⁵ using data rate of 100 Kbps. When the number of PAN source member increases as a result, the contention increases which result in the reduction of the network performance. Meanwhile, at a high data, seriously affected performance results are obtained for the IEEE 802.15.4 because of static BO and SO. However, AAOD¹³ and AMPE¹⁴ only adjust the SD dynamically, which outperform the IEEE 802.15.4. DBSAA¹⁵ is better than the other protocols except CDSA as it adjusts both BO and SO at the same time. Consequently, CDSA outperforms the rest of the algorithms, as the content requirements are known to the PAN, and alters the superframe structure consequently. Hence, CDSA delivers the maximum network throughput when matched to IEEE 802.15.4, AAOD,¹³ AMPE,¹⁴ and DBSAA¹⁵ as shown in Figure 6.

Figure 6.

Average throughput of CDSA and other algorithms against number of nodes at 100 Kbps.

Latency

Figure 7 shows the average end-to-end latency seen at PAN coordinator for CDSA and other algorithms. In this simulation, 10 source nodes are generating overall 250 Kbps of data for the PAN coordinator. In case of IEEE 802.15.4 at BO = 6 and SO = 2, prominent end-to-end latency is observed due to too short active portion to carry all the data packets. Therefore, waiting time for buffered packets increases and results in high end-to-end latency. Dynamic settings of superframe structure by CDSA play an important role in minimizing the end-to-end delay as shown in Figure 7. CDSA accomplishes good performance in terms of end-to-end delay as compared to IEEE 802.15.4 and the other algorithms. Queue size is an important factor that reflects the end-to-end delay. Greater queue length provides higher buffering capacity and increases end-to-end delay.

Figure 7.

End-to-end delay of CDSA and other algorithms.

In Figure 8, the impact of varying queue size on average end-to-end delay is shown. High end-to-end delay is obtained for IEEE 802.15.4, especially when SO is small and queue size is large. This is because the SD is not enough to carry all the packets generated by PAN members and packets stay in the buffer during the inactive duration. Latency is not significantly affected when CDSA is used with large queue size. The reason is that CDSA adjusts its active duration (BO and SO) according to the application requirements and contents, thus allowing packets to stay in the queue.

Figure 8.

Impact of queue length on average end-to-end delay of CDSA and other algorithms.

Energy consumption

Figure 9 shows the percentage of energy consumed during only 500 s of simulation where 10 nodes transmitting data to the PAN coordinator collectively at a data rate of 250 Kbps. It is noticeable that CDSA consumes less energy as compared to AAOD,¹³ AMPE,¹⁴ and DBSAA.¹⁵ The reason is that it adjusts the superframe according to the traffic demands. Moreover, when data reporting stops, it immediately adjusts BO and SO to default values. IEEE 802.15.4 has fixed duty cycle; therefore, SD is half of BI when the value of BI is high at high value of BO. This results in spending more time in inactive Statex by network devices. In other words, the network throughput degrades drastically.

Figure 9.

Percentage of network energy utilized using CDSA and other algorithms.

Conclusion

The dynamic and optimum resource utilization in low-powered devices such as IoMT always remains a challenging issue for research community. A number of dynamic superframe adaptation approaches are developed. However, no such algorithm/approach is available as per literature survey which alters the SD of IEEE 802.15.4 according to the content requirements. In this article, we proposed a novel approach called CDSA algorithm that adjusts the BO and SO dynamically along with BE to balance the trade-off among application requirements in terms of packet deliver ratio, energy consumption, latency of communication and network throughput. CDSA algorithm does not imply any modification to the IEEE 802.15.4. The dynamic approach is adopted by PAN coordinator to adjust the BO and SO. Expected SO is calculated when BI is expired. It adjusts the superframe structure when network load is verified. In the absence of a network activity, CDSA algorithm changes the BO and SO to predetermined settings to conserve the energy. Simulation outcomes indicate that CDSA outperforms IEEE 802.15.4 and other prominent existing algorithms in terms of improving the network performance. This approach is limited for low-powered devices communicating and interacting with each other using IEEE 802.15.4 standard. Therefore, for future perspective, IEEE 802.15.6 standard will be used to implement the application content requirements to enable efficient healthcare paradigm.

Footnotes

Handling Editor: Mohsin Raza

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research,authorship,and/or publication of this article.

Funding

The author(s) received no financial support for the research,authorship,and/or publication of this article.

ORCID iDs

Yousaf Zia

Kashif Naseer Qureshi

References

Qureshi

Bashir

Iqbal

. Cloud computing model for vehicular ad hoc networks. In: Proceedings of the IEEE 7th international conference on cloud networking (CloudNet), Tokyo, Japan, 22–24 October 2018, pp.1–3. New York: IEEE.

Qureshi

Bashir

Abdullah

. An energy and link aware next node selection protocol for body area networks. In: Proceedings of the 2018 international conference on information networking (ICOIN), Chiang Mai, Thailand, 10–12 January 2018, pp.721–726. New York: IEEE.

Azgomi

Jamshidi

. A brief survey on smart community and smart transportation. In: Proceedings of the 2018 IEEE 30th international conference on tools with artificial intelligence (ICTAI), Volos, 5–7 November 2018, pp.932–939. New York: IEEE.

Qureshi

Abdullah

. Adaptation of wireless sensor network in industries and their architecture, standards and applications. World Appl Sci J 2014; 30(10): 1218–1223.

Power

Hadidi

. Transforming agriculture: exploring precision farming research needs. J Midwest Assoc Inform Syst 2019; 2019(2): 1.

Qureshi

Bashir

Islam

. Link aware high data transmission approach for internet of vehicles. In: Proceedings of the 2019 2nd international conference on computer applications information security (ICCAIS), Riyadh, 1–3 May 2019, pp.1–5. New York: IEEE.

Da Xu

Zhao

. 5G internet of things: a survey. J Ind Inform Integr 2018; 10: 1–9.

Qureshi

Bashir

Abdullah

. Distance and signal quality aware next hop selection routing protocol for vehicular ad hoc networks. Neural Comput Appl. Epub ahead of print 25 June 2019. DOI: 10.1007/s00521-019-04320-8.

Holler

Boyle

Karnouskos

, et al. From machine-to-machine to the internet of things. Cambridge: Academic Press, 2014, pp.1–331.

10.

Qureshi

Abdullah

Kaiwartya

, et al. A dynamic congestion control scheme for safety applications in vehicular ad hoc networks. Comput Electr Eng 2018; 72: 774–788.

11.

Qureshi

Abdullah

Bashir

, et al. Cluster-based data dissemination, cluster head formation under sparse, and dense traffic conditions for vehicular ad hoc networks. Int J Commun Syst 2018; 31(8): e3533.

12.

Low-power 24GHz transceiver designed for industrial and consumer IEEE 802.15.4, ZigBee, http://www.atmel.com/devices/at86rf231.aspx

13.

López

Casilari

. An adaptive algorithm to optimize the dynamics of IEEE 802.15.4 networks. In: Proceedings of the Mobile Networks and Management, Würzburg, 23–25 September 2013, pp.136–148. New York: IEEE.

14.

Barbieri

Chiti

Fantacci

. Proposal of an adaptive MAC protocol for efficient IEEE 802.15.4 low power communications. In: Proceedings of the IEEE 49th global telecommunication conference, San Francisco, CA, 27 November–1 December 2006, pp.1–5. New York: IEEE.

15.

Camila

OHS

Yacine

Stephane

. A duty cycle self-adaptation algorithm for the 802.15.4 wireless sensor networks. In: Proceedings of the IEEE transaction on global information infrastructure symposium, Trento, 28–31 October 2013, pp.1–7. New York: IEEE.

16.

Sabita

Pudasaini

Pyun

, et al. A new MAC protocol for emergency handling in wireless body area networks. In: Proceedings of the eighth international conference on ubiquitous and future networks (ICUFN), Vienna, Austria, 5–8 July 2016, pp.588–590. New York: IEEE.

17.

Dahae

Joung

Cho

. Restructing 802.15.4 MAC protocol’s superframe structure for energy and transmission efficiency. In: Proceedings of the 2015 IEEE international SoC design conference (ISOCC), Gyungju, South Korea, 2–5 November 2015, pp.257–258. New York: IEEE.

18.

Hwang

Yundra

, et al. Analysis of superframe duration adjustment scheme for IEEE 802.15.4 networks. EURASIP J Wirel Commun Netw 2015; 2015(1): 1–17.

19.

Qureshi

Abdullah

. Adaptation of wireless sensor network in industries and their architecture, standards and applications. World Appl Sci J 2014; 30(10): 1218–1223.

20.

Nguyen

Khan

Ngo

T. An energy and QoS-aware packet transmission algorithm for IEEE 802.15.4 networks. In: Proceedings of the IEEE 26th annual international symposium on personal, indoor, and mobile radio communications (PIMRC), Hong Kong, China, 30 August–2 September 2015, pp.1255–1260. New York: IEEE.

Content-based dynamic superframe adaptation for Internet of Medical Things

Abstract

Keywords

Introduction

Related work

System model

Proposed methodology

Condition I: SOExp > SOCur

Condition II: SOExp≤ SOCur

Performance evaluation

Packet delivery ratio

Network throughput

Latency

Energy consumption

Conclusion

Footnotes

Declaration of conflicting interests

Funding

ORCID iDs

References

Condition I: SO_Exp > SO_Cur

Condition II: SO_Exp≤ SO_Cur