Energy-aware MAC protocol for data differentiated services in sensor-cloud computing

Abstract

The energy-aware MAC protocol is the basic communication protocol for device-to-device communication in sensor-cloud computing and facilitates data sensing, computing, and sharing for applications. Transmitting high-priority data to control centers quickly to manage emergencies is essential for sensor-cloud applications. In this paper, a Dynamic Wait time-based MAC (DWT-MAC) protocol is proposed for data differentiated services of a sensor-cloud system (SCS). The DWT-MAC protocol is novel in that it changes the receiver wait time, which has been fixed in previous MAC protocols, and it reduces the delay in data transmission by dynamically adjusting the wait time. In the DWT-MAC protocol, the wait time changes according to the number of senders, which can ensure that it always approaches the optimal value. A dynamic time adjustment algorithm, which causes the wait time to always trail the optimal value, is proposed to reduce delays. While the DWT-MAC protocol cannot ensure wait times of the optimal value, it can ensure that it quickly comes close to the optimal value, which is suitable for dynamically changing networks. Extensive experiments show that the DWT-MAC protocol reduces the average delay in the transmission of data of the highest priority by 49.3%.

Introduction

Cloud computing has been widely used by the scientific community and industry, because users can benefit from computing infrastructures at low costs [1,2,3]. At the same time, big data networks have attracted the attention of numerous researchers [4,5,6]. However, the development of big data networks is attributable to the development of numerous applications with several sensing devices [7,8,9]. A sensor-cloud system (SCS) combines cloud computing [10,11,12] with sensor nodes [13,14,15] to sense, collect, process, analyze, store and share data for various applications; this combination has largely promoted the development of big data networks [16,17,18].

In recent years, with the development of microprocessor technologies [19], devices connected to the Internet have become more prevalent [20,21,22] with the number of such devices anticipated to increase to 23 billion by 2020 according to relevant research [19]. The computing, storage, and communication power of these devices has also developed considerably with network edges accumulating numerous resources for computing, storage, and communication, forming the so-called edge network [23,24,25]. This has resulted in a computation shift from the cloud to the edge, and the combination of edge and cloud networks has formed the edge cloud computing mode, which has considerably changed existing networks [26, 27]. However, most of these large devices are embedded sensing devices, which are the main source of big data networks. We can obtain large amounts of data because these sensing devices can sense surrounding environments anytime and anywhere. A Cisco report shows that data flows generated by the Internet of Things (IoT) account for 69% of data flows generated by the Internet, reflecting a 30-fold increase relative to levels in 2000. Data flows are still increasing, continuously promoting the development of big data networks [28]. Data transmission, data protection and intrusion detection technologies can help guarantee the security of IoT data [29].

However, numerous challenges still emerge when massive data enter big data networks. Communication between sensing devices follows a basic protocol [30] and creates challenges in big data networks due to its influence on every aspect of big data network performance [31,32,33,34,35,36]. Device-to-device (D2D) communication has attracted considerable research interest as a basic mode of network computing [30, 36]. The MAC protocol is an important protocol of D2D communication [30, 36]. It influences every aspect of communication performance. Communication performance is also affected by other communication constraints such as bandwidth and communication noise [34,35,36]. The two most important aspects of communication it influences are as follows. (1) Energy efficiency [37]. Generally, most sensing devices are powered by batteries, which provide very limited energy, creating a need for high energy efficiency [38]. Under normal circumstances, node energy consumption is divided into four states as follows: (a) sending data; (b) receiving data; (c) idle listening; and (d) sleeping. The first three states consume the most energy. Therefore, network nodes should sleep when possible to save energy. In most cases, nodes work based on a duty cycle [38], i.e., nodes periodically shift between the active and sleep modes to save energy [39]. (2) Delay in data transmission. Transmitting and storing data reliably and quickly is critical [40]. The delay in data transmission is the expected delay in communication between devices and is mainly attributable to communication protocols [3, 9, 30]. As a node only sends and receives data when it is active, when a node is sleeping, the sender must wait until the receiver wakes for data operation [39], shortening the node’s active time and increasing delays. At the same time, when transmitting many-to-one data through a sensor-cloud, different MAC protocols can greatly influence delays.

The MAC protocol for sensor-cloud systems is divided into two MAC protocols for actively initiating communication: (1) the sender-initiated MAC protocol and (2) receiver-initiated MAC protocol. Under the sender-initiated MAC protocol, when a sender has data to be transmitted, the sender listens to and polls the channel, and if the receiver is awake, the receiver can receive the data transmitted by the sender and establish a communication, resulting in data transmission. Under the receiver-initiated MAC protocol, when the sender must transmit data, it does not initiate data transmission and must wait for a beacon from the receiver indicating that the receiver is ready to receive data [41]. When the receiver wakes, it actively sends the beacon to receive data and initiate communication. As the sender waits for the beacon, it can communicate with other receivers until achieving data transmission [41].

In most sensor-cloud systems, sensing devices are relatively dense, and there are always multiple sensing devices with data to transmit at the same time within a communication range; in other words, there are a many-to-one cases of communication in which multiple senders must transmit data to the same receiver [39]. Data from different senders usually is of varying priority [41], and of course high priority data must be transmitted first, as this is good for the system [41]. For example, in an emergency, alarm data should be transmitted first while less important data such as daily life data can be transmitted later. Therefore, researchers have designed data transmission protocols that prioritize data in the MAC protocol layer. Such protocols are helpful, as in many cases of sensor-cloud system communication, there is only one channel, meaning that high priority data must be transmitted first when multiple senders must transmit data. Sarang et [41] developed an MAC protocol allowing a receiver-initiated network to transmit high priority data faster. Under this protocol, when the receiver wakes up, it broadcasts a wake up beacon to indicate that it is ready for data transmission; then, it initiates a wait time of a fixed duration for sender nodes to reply to it. The wait time is usually determined by the number of sender nodes. Thus, if there are n sender nodes, the receiver’s wait time is equal to the duration required to transmit n Tx-Beacons. If the sender node has data to transmit, it will listen to the channel. If it receives a wake up Beacon from the receiver, it transmits a Tx-beacon to the receiver. Because a Tx-Beacon contains priority data, after the receiver receives Tx-Beacons from all sender nodes, it communicates with the sender with data of the highest priority to transmit this information faster.

While past studies have realized ways of transmitting prioritized data, such methods present the following shortcomings: (1) Long delays. When there are numerous sender nodes, wait times are long, causing establishing communication before data transmission to take a considerable amount of time for each set of data and resulting in long delays. In a multi-hop network, multi-hop data transmission results in considerable delays. (2) Inefficient energy consumption. While there are many sender nodes in a network, not every sender node has data to transmit most of the time. However, a receiver must reserve time to transmit Tx beacons for each sender node. When some sender nodes do not have data to transmit, some wait time is wasted because the receiver must reserve time for these sender nodes, resulting in long periods of empty consumption, invalid energy consumption, and long network delays.

The above analysis shows that the current MAC protocol for prioritized data is mainly limited in that the wait time set by the receiver is not equal to the number of sender nodes, and the fixed wait time results in long delays and low energy efficiency. This is the case because while a receiver may have multiple sender nodes, not every sender node has data to transmit. The method reserves time for each sender node, which limits the protocol’s flexibility and performance. Thus, a more energy efficient Dynamic Wait time-based MAC (DWT-MAC) protocol with shorter delays is proposed in this work. The paper makes the following contributions:

• (1) In this paper, a Dynamic Wait time based MAC protocol is proposed to reduce delays and enhance the energy efficiency of data differentiated services of a sensor-cloud system (SCS). The DWT-MAC protocol differs from the previous strategy with a fixed wait time by reducing data transmission delays by dynamically adjusting the wait time. Under the DWT-MAC protocol, wait times change depending on the number of senders, which can ensure that they are always close to optimal. This dynamic wait time strategy with wait times always close to the optimal value is intended to reduce delays. While the protocol cannot ensure that wait times are always of the optimal value, it ensures that wait times are close to the optimal value, making it very suitable for networks with changing data.

• (2) Our numerous studies and experiments show that the strategy proposed here is better than the previous strategy. Under the DWT-MAC protocol, when wait times exceed the number of senders (i.e., for f senders, the wait time exceeds the period required to transmit f Tx-Beacon), this has almost no effect on high priority data. When wait times fall below what is required for the number of senders (i.e., wait times fall below what is required to transmit f Tx-Beacon), this may influence some high priority data, but network data transmission delays can be reduced. In both cases, the wait time is adjusted before the next round of data transmission, causing wait times to approach optimization. Our numerous experiments show that the DWT-MAC protocol is better than the previous strategy by 49.3%.

The rest of this paper is organized as follows. In “Related work” section, we review the related work. “System model and problem statement” section describes the network model and formulate the optimization problem to reduce delay. In “The design of the DWT-MAC protocol” section, we propose DWT-MAC protocol. Simulation results are presented in “Experimental results and analysis” section and the entire paper is concluded in “Conclusion” section.

Related work

With the development of the Internet of Things (IoT) [15], more and more sensor devices are deployed in various applications, such as vehicles network [18, 42], traffic monitoring, healthcare [43], security monitoring [44,45,46], and industrial networks [47,48,49]). Numerous sensor devices are used to sense and collect data, providing rich data for various applications [50]. As the computing and storage power of these devices has improved and with the development of microprocessor technology, a major shift in current network architectures has occurred. The large number of computing and storing resources gathered at the edges of networks have formed Edge networks. The center of network computing has thus shifted from the cloud to the edge. Edge networks generate large amounts of data and have formed big data networks. The combination of these forms has led to a considerable shift in network computing modes. Networks are developing from those of single cloud computing those combining edge and cloud computing [1, 2, 5, 7].

Among sensor devices, device-to-device communication plays an important role in network performance. Data transmission delays and network energy consumption are two main issues of concern, and the basic MAC protocol plays an important role in network performance. This section introduces research conducted in this area.

MAC protocol

The MAC protocol is vital for device-to-device communication and directly influences network performance through energy consumption, delays, etc. [3, 9, 30]. The protocol has attracted considerable attention due to its importance. Past works related to the research presented this paper is introduced below.

In most networks, the time ratio for devices used in data operations is not high. Therefore, most devices use a duty cycle scheme, which causes devices to periodically alternate between the awake and sleep modes. As devices consume 1/1000 of their energy capacity when asleep, a duty cycle scheme can save considerable amounts of energy [39]. However, while a duty cycle scheme can increase network lifetimes, it can have negative impacts on other areas of network performance. Its influence on delays is the most significant. In a network that does not use a duty cycle, multiple senders must transmit data at the same time, increasing delays. In a network that uses the duty cycle, when a sender must transmit data, it must wait until the receiver wakes, also increasing delays. The Sensor-MAC (S-MAC) protocol employs a duty cycle based the MAC protocol [51]. Under the S-MAC protocol, while nodes save energy by periodically waking and sleeping, even when there is no data operation, the amount of time spent awake is stationary. Therefore, researchers have proposed the Time out MAC (T-MAC) protocol [52], which keeps time spent awake constant even during times without data operation under the S-MAC protocol. A node listens the channel for a period of time, and when there is no data operation, it goes to sleep to save energy, increasing delays. Under duty cycle-based MAC protocol, the length of a duty cycle influences delays most. Usually, the proportion of awake time is greater and delays are shorter, but energy consumption is higher [52,53,54,55].

The above analysis is based on the contention-based MAC protocol, which requires devices to be periodically awake even when there is no data operation. The Time Division Multiple Access (TDMA) protocol [56] determines device work slots in advance. Devices are awake during data operation and go to sleep when data operation is complete, saving energy. However, this MAC protocol must plan work slots for each device beforehand. Therefore, it is not suitable for dynamically changing networks [56].

Under the divided MAC, receiver-initiated MAC and sender-initiated MAC protocols, one receiver and one sender form the proactive side of data transmission [41]. The receiver-initiated and sender-initiated MAC protocols are similar and differ only in the timing of data operation. Under the sender-initiated MAC protocol, when a sender has data to transmit, the sender uses to non-stop polling channel, and the receiver listens to the channel after it wakes; if the receiver finds a sender polling channel, it can communicate with the sender. It is apparent that under this MAC protocol, both the polling frequency of the sender and the time required for the receiver to listen to the channel will affect data transmission delays [41]. When the polling channel frequency of the sender is high, the receiver receives polling signals easier after it wakes, causing it to communicate with the sender earlier and shorten data transmission delays. However, the energy consumed by the sender increases. The amount of time required for the receiver to listen to the channel also affects delays, because when this period is short, the receiver will go to sleep after listening; when the sender has data to transmit in this period and a data request must be sent, it must wait until the receiver wakes, resulting in long delays.

On the other hand, under the receiver-initiated MAC protocol, when a sender has data to transmit, it stays silent and listens to the channel. The receiver broadcasts a wake up beacon showing that it is ready to receive data transmission requests from senders after wakes. Therefore, after the sender receives the wake up beacon from the receiver, the sender sends a Tx-Beacon requesting data transmission. Thus, under the receiver-initiated MAC protocol, the receiver’s wake up-Beacon influences delays most. When a receiver only needs to send a wake up-Beacon once, a sender waking before the receiver sends a wake up-Beacon can receive the wake up-Beacon and thus initiate data communication. Senders waking after the receiver sends a wake up-Beacon and wanting to establish data communication must wait for the next cycle, increasing delays. When a receiver intermittently sends a wake up-Beacon after it wakes, this shortens delays but requires more energy consumption.

Priority data transmission

As delay sensitivity varies for different perceived data, when large amounts of data must be transmitted at the same time, prioritizing emergency data transmission is essential for applications. For example, emergency data related to fires, alerts and etc. must be transmitted as soon as possible so that the decision center can address the emergency quickly. Some works have thus focused on priority data transmission. Priority data transmission can be divided into the following types based on different methods used:

• (1) Priority data transmission in the MAC layer. Priority data transmission based on the MAC layer can transmit high priority data to the channel first in the bottom layer of the network or transmit high priority data after determining priorities, lowering system costs and improving system performance. The contention window is the amount of time required for random backoff when multiple devices are using a channel; devices randomly select a period of time to wait for and listen to the channel, and data transmission starts if the channel is still idle after this period of time [57]. The length of the contention window is critical to the contending channel. Generally speaking, a contention window has following effects on devices using a channel. When a contention window is large, in terms of probability, devices must wait a long time, so it a device selects a shorter contention window, shortening the wait time, the probability of winning the competition for devices selecting a shorter contention window is higher, increasing the likelihood of data from these devices being transmitted first [57]. Obviously, devices use smaller contention windows for high priority data and larger ones for low priority data. Therefore, high priority data are likely to be successful in the contending channel, ensuring that high priority data are sent first. Setting a smaller contention window not only ensures that high priority data are sent first, but also reduces delays because the contending channel uses less time, reducing energy consumption and increasing network lifetimes. However, when the size of the contention window is reduced, there will be more collisions channels compete, degrading system performance. Priority data transmission that involves setting the size of the contention window is most advantageous due to its low costs. With this method, the MAC protocol need not pay extra costs for priority data transmission. It only needs to change the size of the contention window. However, for this method, data priorities are undetermined, meaning that it is impossible to determine which data are transmitted first when there are multiple senders. This the case because even when the contention window is small, the wait time randomly selected by devices with a small contention window may still be shorter than that chosen by devices with a large contention window, and thus devices with lower data priority access the channel earlier. Therefore, this priority data transmission method is effective, but it cannot guarantee that high priority data are transmitted first.

S. Sarang et al. [41] proposed another reliable strategy for priority data transmission where high priority data are transmitted first. In a receiver-initiated network, after a receiver wakes, it broadcasts a wake up-Beacon which indicates that a sender with data to transmit can send a request to transmit data. A sender sends a Tx-Beacon to request data transmission, and priority data it wants to transmit is contained in the Tx-Beacon. The receiver returns the Rx-Beacon to select the sender with the highest priority data after it receives Tx-Beacons from all senders, and thus devices with high priority data can transmit data first. This method guarantees that high priority data are transmitted first. However, the system’s costs are high, as a period of time is set for each Tx-Beacon.

• (2) Priority data transmission in the network routing layer. Priority data transmission based on a network routing layer is another commonly used and diverse method. The priority queue method is frequently used for priority data transmission and is based on the network routing layer. The method sets multiple queues by data priority, and senders choose data to send by polling when they can send. Polling starts with the highest priority queue, and if there are high priority data to transmit, these data are selected and sent; otherwise, the lower priority queue is queried. Using this method can guarantee that high priority data are transmitted first. However, while the sender can ensure that its high priority data are transmitted first, it cannot guarantee that its data will be sent. When a sender cannot send data, its high priority data cannot be sent, limiting this method.

In the routing layer, data transmission delays not only affect delays in each hop of data transmission, but also the number of hops required for routing. If the number of hops needed for data transmission routing is large, delays will be long as well. Thus, a routing path with few hops should be used for emergency data. Interference emergency data routing should be avoided by choosing the path with more routing hops for data that are not urgent. To select different routing paths based on data priority, Zhang et al. [58] proposed the integrity and delay differentiated routing (IDDR) strategy. Under the IDDR scheme, delay-sensitive data choose the shortest routing path for data transmission to guarantee delay minimization. Longer routing paths for data that are delay-insensitive but with high integrity requirements are selected to prevent the shortest routing path from being chosen by emergency data. Thus, integrity is guaranteed, and network performance and data transmission with different requirements can be ensured by selecting different routing paths.

System model and problem statement

The network model

The network model used in this paper is similar to that used in Ref. [41] as is described as follows: a wireless sensor network contains a sink node and multiple ordinary nodes with the same characteristic V = {v₁, v₂, ……, v_{(n − 1)}, v_n}; for ease of expression, the network described in this paper is abstracted to a plane tree network, and each ordinary node in the tree network has sole parent nodes but can have multiple child nodes. A tree network with n ordinary nodes can be expressed as W = {sink, v₁, v₂, ……, v_{(n − 1)}, v_n}, and the sink node is expressed as the root node in the tree network. Each node in the sensor network can sense its surrounding environment and can send sensing data to the node in the next hop. The next hop node makes fewer hops to the sink node, and the node sends data to the sink node by multi-hop routing. Ordinary nodes in the network are powered by batteries of the same limited capacity, and they can harvest energy from the surrounding environment; the energy of the sink node is infinite.

Data priority is divided to four levels: P₄ (emergency), P₃ (most important), P₂ (important), and P₁ (not important). P₄ denotes the highest level of data priority. Data in the network are expressed as P_ack, and data of the highest priority in the network are expressed as P_hi. P_hi ⊆ P_ack.

Some symbols used in this paper and their definitions are given in Table 1.

Table 1 Symbols used in this paper and their definitions

The problem statements

Definition 1

Packet delay. The time required for the data is successfully transmitted to the sink node. If a data ${P}_{ac}^i$ is generated at time ${T}_g^i$, it is successfully transmitted to the sink node at time ${T}_e^i$. The packet delay ${T}_t^i$ of these data can be expressed as

$$ {T}_t^i={T}_e^i-{T}_g^i $$

(1)

Definition 2

Total delay. The sum of the packet delay of multiple data. If there are x data in the network, the total delay T_to of these x data can be expressed as

$$ {T}_{to}={\sum}_{i=1}^x{T}_t^i $$

(2)

In the network, the total delay should be as low as possible to ensure the data can be transmitted in time and processed quickly. Thus, the first focus of this paper can be expressed as

$$ \left\{\begin{array}{c}\mathit{\min}\left({T}_{to}\right)=\mathit{\min}\left(\sum {T}_t^i\right)\mid i\in {P}_{ack}\ \\ {}s.t.{T}_t^i\ne 0\mid i\in {P}_{ack}\ \end{array}\right. $$

(3)

Definition 3

Average highest priority packet delay. The average packet delay for multiple data of the highest priority. If there are x data of the highest priority in the network, the average delay of highest priority packets T_ave of these x data can be expressed as

$$ {T}_{ave}=\left({\sum}_{i=1}^x{T}_t^i\right)/x $$

(4)

In the network, the average delay of highest priority packets should be as low as possible to ensure that the highest priority data can be transmitted as soon as possible and so that emergencies can be addressed earlier. Thus, the second goal of this paper can be expressed as

$$ \left\{\begin{array}{c}\mathit{\min}\left({T}_{ave}\right)=\mathit{\min}\left(\frac{\sum {T}_t^i}{N_{hi}}\right)=\mathit{\min}\left(\sum \frac{T_e^i-{T}_g^i}{N_{hi}}\right)\mid i\in {P}_{hi}\\ {}s.t.{T}_t^i\ne 0\mid i\in {P}_{hi},{N}_{hi}\ne 0,{P}_{hi}\ne \varnothing \end{array}\ \right. $$

(5)

Overall, we seek to minimize the total delay, D, and the average delay of highest priority packets, A, which can be summarized as follows:

$$ \left\{\begin{array}{c}\mathit{\min}(D)=\mathit{\min}\left({\sum}_{i\in {P}_{ack}}\Big({T}_e^i-{T}_g^i\right)\\ {}\mathit{\min}(A)=\mathit{\min}\left({\sum}_{i\in {P}_{hi}}\frac{T_e^i-{T}_g^i}{N_{hi}}\right)\end{array}\right. $$

(6)

The design of the DWT-MAC protocol

In improving the previous receiver-initiated MAC protocol, we first introduce the QoS MAC protocol for Prioritized Data (QPPD-MAC) below and improve it. After discussing QPPD-MAC protocol communication, our proposed protocol is given.

Background information

Communication under MAC protocols for prioritized data such as the QPPD-MAC is illustrated in Fig. 1. The network shown in Fig. 1 is a many-to-one data transmission network with three sender nodes that want to transmit data to the same receiver. Nodes in the network with data to transmit wake and listen to the channel. The receiver performs clear-channel assessment (CCA) once it wakes to determine the status of the medium; if the medium is idle, it broadcasts a wake up-Beacon indicating that it is ready to receive data. The wake up-Beacon contains a Source Address (SA) and Energy State (ES) as shown as Fig. 2. The receiver then initiates a wait time of duration T_w to receive Tx-Beacons from the sender nodes, and the Tx-Beacon contains an SA, Destination-Address (DA), Priority (PR), and Network-Allocation Vector (NAV) (see Fig. 2). The duration of wait time T_w is proportional to the number of sender nodes $\mathcal{N}$, and T_w not only contains the time length for CCA but also the time length for the $\mathcal{N}$ Tx-Beacon. After sender nodes receive the wake up-Beacon from the receiver, they send Tx-Beacons to the receiver after they successfully compete for the channel. The receiver broadcasts the Rx-Beacon with the address of the selected sender when one of the following two conditions is met. The Rx-Beacon contains an SA, Selected Node (SN), and NAV (see Fig. 2). The two conditions are as follows:

• (1) the timer expires and the wait time reaches the predetermined value T_w;

• (2) Timer cancelled, receiver received the Tx-Beacon with the highest data priority. Sender nodes compare the SN contained in Rx-Beacon to itself address after successfully received Rx-Beacon, if the SN is same as itself address, this indicates that this sender node is selected for data transmission, and this sender node transmits data packet after a Short Interframe Space (SIFS), finally receiver reply a ACK packet to the selected sender after successfully received data packet. The above process completes the data transmission of a data. Know from the process, when there are multiple sender nodes which have data need to be transmitted, it can guarantee the data which obtain the chance for data transmission must be the data with the relative highest priority among all sender nodes, and finally this data was transmitted to receiver. The subsequent data transmission process is same as above process. The different just is there is no data to be transmitted in sender 1, so sender 1 does not send Tx-Beacon no longer, only senders which have data need to be transmitted still send Tx-Beacon. But receiver did not know the sender 1 has no data to transmit, T_w remains unchanged. The transmission process is shown in Fig. 1. The frame formats of beacons in the transmission process are shown as Fig. 2.

Fig. 1

Data transmission process of the QPPD-MAC protocol

Fig. 2

Frame format of beacons

Research motivation

From the MAC protocol for prioritized data, to transmit high priority data first, it takes a long period of T_w to determine what data to receive from which sender. A time length of T_w can ensure that each sender sends a Tx-Beacon for data transmission, ensuring that the receiver selects the sender with the highest priority data among $\mathcal{N}$ senders. This method guarantees that high priority data are transmitted first, but it has the following limitations:

• (1) Long data transmission delays. The previous MAC protocol based on data priority is similar in that time length T_w is equal to the number of sender nodes when there are multiple sender nodes. This requires data to take a long time to establish communication before data transmission, increasing delays.

• (2) Ineffective energy consumption. As mentioned above, because a receiver needs set a period of time for each sender node, it waits for sender nodes sends a Tx-Beacon in this period of time. However, while there may be multiple sender nodes in a network, most of the time, only some sender nodes have data to transmit, resulting in long periods of empty consumption, ineffective energy consumption, and long delays.

We use a concrete example to analyze the impact of T_w on network performance. For ease of analysis, for networks described in section 4, assume that no new data are generated after data transmission in each cycle. The index of node v_c in the network (i.e., c) is larger, and this node can compete for the channel faster.

The MAC protocol proposed in this paper focuses on the time used for high priority data, and the transmission process is the same after communication is established. Therefore, to make the transmission process shorter and clearer, it is simplified as shown in Fig. 3. Transmission from the receiver broadcasts a wake up-Beacon until the receiver initiates the wait time through a simplified Wake up (Wu) process. The transmission process whereby the receiver broadcasts an Rx-Beacon until data are successfully transmitted to the receiver is simplified as the Data transmit (Dt) process.

Fig. 3

Simplify the data transmission process

To accurately compare the optimization of T_w under different MAC protocols, we do not use time units to represent delays and instead use the number of cycles and the required number of Tx-Beacons to represent delays, which is more accurate. A delay in this paper is defined as time used for multiple cycles. The basic transmission process is generally the same for each data transmission process. A node wakes, a receiver sends a wake up-Beacon, the receiver sends a Rx-Beacon, the sender node sends a data packet and the receiver returns a ACK packet. The time required for each cycle is the same. The cycles differ in transmission processes required for receivers to initiate wait times and receive Tx-Beacons. The wait time stops when the time expires or is cancelled, and the timer is stopped depending on how many Tx-Beacons are transmitted, resulting in different periods of data transmission between cycles. To facilitate comparisons, the time required for the current cycle can be expressed as the total number of cycles passed and the total number of Tx-beacons transmitted. The current time T_cur can be expressed as:

$$ {T}_{cur}= yT+{zT}_{tx} $$

y indicates the total number of cycles passed until the current cycle is reached, and z denotes the total number of Tx-Beacons received by the receiver until the current cycle is reached. We provide an example to explain our delay calculations. For the data transmission process shown in Fig. 4, using the above delay calculation method, for data generated at time 0T + 0T_tx, a sender first sends a Tx-Beacon for the highest priority data. Thus, the receiver’s wait time is cancelled early. When the receiver successfully receives data of the highest priority, the network passes a cycle, and the receiver receives 1 Tx-Beacon. The total delay of this transmission process is expressed as 1T + 1T_tx. The packet delay of the data of the highest priority is 1T + 1T_tx, as only one set of data of the highest priority is successfully transmitted. The average delay of highest priority packets is expressed as 1T + 1T_tx.

Fig. 4

An example of data transmission

In the following, we analyze limitations of the previous MAC protocol and describe the DWT-MAC protocol proposed here. We use network 1 as an example to illustrate limitations of the previous MAC protocol. An initial topology diagram of network 1 is shown in Fig. 5, and the priority levels of data from each node in network 1 are listed in Table 2. Delays of the three cases experienced when the length of T_w is fixed are analyzed to illustrate limitations of the previous MAC protocol.

Fig. 5

Initial topology diagram of network 1

Table 2 The priority levels of data of each node in network 1

• (1) The length of T_w is equal to the number of sender nodes. In network 1 shown in Fig. 5, when T_w = 3T_tx, during the transmission of data from v₁ and v₂ to the sink node, v₁ successfully transmits its data of the highest priority to the sink node in the first cycle, and only v₂ still has data to transmit in the next 2 cycles. Thus, just one sender node v₂ will wake and transmit data in the second and third cycles. As shown in Fig. 6, in these 2 cycles, T_w must be long enough to transmit one Tx-Beacon, but T_w must be long enough to transmit three Tx-Beacons, as no high priority data must be transmitted. The wait time is thus not cancelled early, and the receiver must wait for the wait time of duration T_w to expire, thus wasting time while waiting to receive a Tx-Beacon (see the red lines in Fig. 6). In this period, the sink node and v₂ are waiting. This increases the average and total delays of highest priority packets, network energy consumption increases and the network lifetime is shortened.

• (2) The length of T_w is lower than the number of sender nodes, and assuming T_w = 1T_tx, it loses the feature of priority data transmission. While network data are still prioritized, the receiver only can receive one Tx-Beacon from the sender node at a time; in fact, the receiver starts data transmission after the first Tx-Beacon is received in each cycle. The transmission process at this time is more consistent with the transmission process for unprioritized data than the common protocol, and the QPPD-MAC protocol must transmit a Tx-Beacon, Rx-Beacon and others. At this point, the QPPD-MAC cannot ensure that high priority data are transmitted first, and thus it cannot reduce the average delay of highest priority packets. Instead, the average and total delays of the highest priority packets increase, and network energy consumption increases as well.

• (3) Finally, when T_w = 2T_tx (i.e., the length of T_w is equal to the number of sender nodes), multiple nodes wake up at the same time, v₁ completes data transmission in the first cycle, and after this cycle, when v₂ transmits data in the next 2 cycles, some time is wasted while waiting to receive a Tx-Beacon. While less time is wasted than when T_w = 3T_tx, invalid waiting time is unavoidable.

Fig. 6

The transmission process when T_w = 3T_tx in network 1

Our analysis of the above three cases shows that whatever the length of T_w, it will influence delays and energy consumption. If T_w is too long, high-priority data are transmitted first, but delays increase significantly. When T_w is too short, delays decrease, but the transmission of high priority data first cannot be guaranteed.

Thus, is it possible to develop a method that transmits high priority data as early as possible while limiting delays?

Under the previous MAC protocol, the wait time is fixed, and its length T_w does not change with the state of the network, resulting in high network energy consumption and long delays. When there are k sender nodes with data to transmit, if the value of T_w can be adjusted to T_w = kT_tx before the timer starts, this value can ensure that all sender nodes participate in transmission and that sender nodes with the highest priority data send Tx-beacons in this transmission cycle. This also ensures that T_w is not too large, shortening invalid wait times and solving the above problem. If the value of T_w in each cycle can be adjusted to the optimal value kT_tx, the desired goal can be met. However, because Tx-Beacons are transmitted after the wait time is initiated, the value of T_w cannot be adjusted according to the number of Tx-Beacons received in the current cycle and can only be adjusted by estimating how many Tx-Beacons may be received in the current cycle according the number of Tx-Beacons received in the previous cycle.

To our knowledge, no MAC protocol can dynamically adjust wait times according to the number of Tx-Beacons successfully received. Therefore, there is a need for an MAC protocol that can reduce delays by dynamically adjusting wait times according to the number of Tx-Beacons successfully received. A novel dynamic timer adjustment algorithm is proposed in section 4.3, and an MAC protocol is proposed based on this algorithm.

DWT-MAC protocol

As the wait time is fixed throughout transmission in the previous QoS MAC protocol, a novel Dynamic Wait time based MAC (DWT-MAC) protocol is proposed in this section based on the ideas analyzed in section 4.2. Under the DWT-MAC protocol, the wait time is adjusted according the following rules after the wait period stops in each cycle to bring the value of T_w as close to the optimal value as possible before the next cycle:

• (1) The initiated value of T_w must be set before transmission starts, T_w = nT_tx.

• (2) If the wait time is stopped by the timer in a cycle, do not change the value of T_w in this cycle, and the wait time stays at its original time length.

• (3) If the wait time is expires in a cycle, the current wait time is T_w = iT_tx, and thus count the total number of Tx-beacons successfully transmitted in this cycle as j. If j < i, adjust the value of T_w to T_w = jT_tx, and if j = i, adjust the value of T_w to T_w = (j + 1)T_tx.

• (4) If an exception occurs during transmission and data transmission is not completed, do not change the value of T_w in this cycle, and thus keep the wait time at its original length.

A dynamic timer adjustment algorithm is designed based on the above rules, and the algorithm is written as Algorithm 1. Symbols used in Algorithm 1 and their definitions are given in Table 3.

Table 3 Symbols used in Algorithm 1 and their definitions

Next, in comparing the data transmission processes of the DWT-MAC and previous MAC protocols, how the protocol proposed in this paper reduces the average delay of highest priority packets and total delays by dynamically adjusting the wait time is analyzed. In the following, the transmission process of the other MAC protocol is called the common MAC protocol.

The network for comparing network 2 (the initial topology diagram is shown as Fig. 7) and the priority of data in each node in network 2 is listed in Table 4.

Fig. 7

Initial topology diagram of network 2

Table 4 The prioritization of data of each node in network 2

A case in which the initial wait time is short is analyzed first, setting the initial value of T_w as T_w = 3T_tx. The transmission process of this initial wait time is shown in Fig. 8.

Fig. 8

In network 2, the initial wait time is T_w = 3T_x during transmission under the common MAC protocol

From the transmission process of the MAC protocol, in the first 2 cycles, v₄ cannot transmit data of the highest priority. As T_w cannot change under the common MAC protocol until third cycle v₁ has successfully transmitted all data it contains to the sink node, v₄ can only transmit its Tx-Beacon of the highest priority data to the receiver and participate in data transmission. Sender nodes in the network continuously generate new data, preventing v₄ from transmitting a Tx-Beacon to the receiver and leading to a very long delay for these high-priority data. Sometimes, such data may not be transmitted at all. The packet delay of these data is 3T + 9T_tx, as there is just one high priority dataset and the average delay of highest priority packets is thus 3T + 9T_tx.

For the same network (i.e., in network 2, the initial value of T_w is T_w = 3T_tx), the transmission process that occurs until the highest priority data are transmitted to the sink node is shown in Fig. 9.

Fig. 9

In network 2, the initial wait time is T_w = 3T_x during transmission under the DWT-MAC protocol

Under the transmission process of the DWT-MAC protocol, T_w is dynamically adjusted in each cycle according the dynamic timer adjustment algorithm. The wait time is stopped once the timer completes the first cycle, and the number of Tx-Beacons received by the sink node within T_w is i = 3. The wait time was T_w = 3T_tx, so j = 3 and i = j = 3 at this time, and thus the value of T_w is adjusted to T_w = 4T_tx after the wait time ends in the first cycle. Thus, all four sender nodes can transmit their Tx-Beacons in the second cycle, and v₄ is transmitted the highest priority data in the second cycle, and the highest priority data are successfully transmitted to the sink node early. The packet delay of this highest priority data and of high priority packets is 2T + 7T_tx at this point.

Relative to the common MAC protocol, the DWT-MAC protocol decreases the average delay of high priority packets as follows

$$ 3T+9{T}_{tx}-2T-7{T}_{tx}=1T+2{T}_{tx} $$

In network 2, when the initial wait time is long, the common MAC protocol generates longer wait times, causing sender and receiver nodes to waste time waiting. How the DWT-MAC protocol reduces delays by dynamically adjusting the wait times is analyzed next. We set the initial value of T_w as T_w = 5T_tx, and the first 3 cycles of the common MAC protocol are shown in Fig. 10.

Fig. 10

In network 2, the initial wait time is T_w = 5T_x under the transmission process of the common MAC protocol

Under the common MAC protocol, in the second and third cycles, after the sink node wakes, as the timer sets a duration of T_w = 5T_tx while only three sender nodes must transmit data, time is wasted after the third Tx-Beacon is transmitted, and sink and sender nodes must wait for the wait time to end, increasing packet delays and wasting energy. The total delay of the first cycles is 3T + 14T_tx.

For the same network state (i.e., in network 2, the initial value of T_w is T_w = 5T_tx) and DWT-MAC protocol, the transmission process is shown in Fig. 11.

Fig. 11

In network 2, the initial wait time is T_w = 5T_x under the transmission process of the DWT-MAC protocol

The first 2 cycles of the DWT-MAC protocol are the same as the common MAC protocol, but in the second cycle, the wait time stops once the timer ends, the number of Tx-Beacons received by the sink node is j = 3, and the wait time is T_w = 5T_tx. Thus, i = 5 and j < i, so T_w is adjusted to T_w = 3T_tx after the wait time ends in the second cycle. Thus, in the third cycle, when there are three sender nodes in the network, the wait time is just long enough to receive the Tx-Beacons transmitted by these three sender nodes, and no time is wasted waiting. The total delay of the first 3 cycles is 3T + 12T_tx. In this network state, relative to the common MAC protocol, the DWT-MAC protocol reduces total delays by

$$ 3T+14{T}_{tx}-3T-12{T}_{tx}=2{T}_{tx} $$

From above comparisons, the DWT-MAC protocol can reduce the average delay of high priority packets and total delays by dynamically adjusting the wait time.

Experimental results and analysis

In this section, we compare the DWT-MAC protocol to the QPPD-MAC protocol based on a series of simulations. In the simulations, the transmission failure rate is set to 1/10000, and we calculate the total delay and the average delay of highest priority packets. The DWT-MAC and QPPD-MAC protocols mainly differ based on whether wait times can be dynamically adjusted in each cycle. The performance of the DWT-MAC protocol is evaluated through these simulations.

The network topology of our simulations has two levels: the sink node and ordinary nodes. When the number of ordinary nodes in network is l (i.e., W = {sink, v₁, v₂, ……, v_{(l − 1)}, v_l}), the topology of this network is as shown in Fig. 12.

Fig. 12

Network topology diagram of simulations

Our simulation proceeds as follows:

We first determine if the current cycle must generate new data from the minimum value of total data (Mn). If the current volume of data (Cn) is equal to or greater than Mn, no new data need be generated, and otherwise new data must be generated one by one until Cn is equal to Mn. Which ordinary node generates data and how the data are prioritized as randomly determined. Then, each sender node generates a random value to compete for the medium. Sender nodes transmit Tx-Beacons in descending order of these random values until the wait time is stopped by timer expiry or cancelling. After transmitting Tx-Beacons, the sender node with the highest priority data transmits data to the sink node.

Our simulations compare the DWT-MAC protocol to the QPPD-MAC protocol based on different volumes of data in each cycle, different numbers of ordinary nodes in each network and different changes in the total volume of data. The period is set to T_w = 3T_tx in each simulation. The simulations calculate the total and average delays of high priority packets for the first 5000 cycles. Because the number of cycles passed under the DWT-MAC protocol is always the same as the number of cycles passed under the QPPD-MAC protocol, when comparing total delays, only wait times are compared.

Symbols used in the simulations and their definitions are shown in Table 5.

Table 5 Symbols used in simulations and their definitions

Delay comparisons

The total number of periodic data changes

This section shows what results from periodic data changes in each cycle. The transmission delays of the DWT-MAC and QPPD-MAC protocols are evaluated under different conditions.

When the total number of periodic data changes shifts from 0 to 6, Mn and Cn for the first 50 cycles are shown in Fig. 13. The average delay of high priority packets is shown in Fig. 14. At this point, despite the total number of ordinary nodes, the average delay of highest priority packets is reduced under the DWT-MAC protocol. When the total number of ordinary nodes is 6, 12 or 18, the average delay of high priority packets for the DWT-MAC protocol respectively decreases by 3.6%, 8% or 9.6% relative to the QPPD-MAC protocol. The average delay of highest priority packets is lower for the DWT-MAC protocol, but the effect is not significant, with the total number of ordinary nodes increasing and the average delay of high priority packets decreasing more. The total delay for this case is shown in Fig. 15. From the results, the total delay of the DWT-MAC protocol is increased despite the total number of ordinary nodes relative to the QPPD-MAC protocol. As the number of ordinary nodes increases, the total transmission delay increases more, demonstrating a different trend from the average delay of highest priority packets.

Fig. 13

The total number of periodic data changes increases from 0 to 6; Mn and Cn in the first 50 cycles

Fig. 14

The total number of periodic data changes increases from 0 to 6; the average delay of highest priority packets

Fig. 15

The total number of periodic data changes increases from 0 to 6; the total delay

As data in the network increase, the value periodically changes from 0 to 12.

Mn and Cn for the first 50 cycles are shown in Fig. 16. Average delays of high priority packets for different total numbers of ordinary nodes are shown in Fig. 17. In this case, when the number of ordinary nodes is 6, 12 or 18, the average delay of high priority packets for the DWT-MAC protocol is decreased by 22.4%, 35% or 37.8%, respectively, relative to the QPPD-MAC protocol. The average delay of high priority packets still decreases despite the total number of ordinary nodes, but the reduction is significant. As the volume of data periodically changes from 0 to 6, the number of ordinary nodes increases, and the average delay of high priority packets is reduced more. The total delays of the DWT-MAC and QPPD-MAC protocols are shown in Fig. 18. The total transmission delay of the DWT-MAC protocol is higher than that of the QPPD-MAC protocol, and as the number of ordinary nodes increases, the total transmission delay increases more. Thus, the average delay of high priority packets is decreased more, and the total delay also increases more in this case.

Fig. 16

The total number of periodic data changes increases from 0 to 12; Mn and Cn in the first 50 cycles

Fig. 17

The total number of periodic data changes increases from 0 to 12; the average delay of highest priority packets

Fig. 18

The total number of periodic data changes increases from 0 to 12; the total delay

The change in the total volume of data increases more when it periodically changes from 0 to 18. Mn and Cn for the first 50 cycles are shown in Fig. 19. The average delays of high priority packets under the DWT-MAC and QPPD-MAC protocols are shown in Fig. 20. In this case, when the number of ordinary nodes is 6, 12 or 18, the average delay of high priority packets under the DWT-MAC protocol is respectively decreased by 33.4%, 47% or 50% relative to the QPPD-MAC protocol. The total delays of the DWT-MAC and QPPD-MAC protocols are shown in Fig. 21, and that of the former increases more.

Fig. 19

The total number of periodic data changes increases from 0 to 18; Mn and Cn in the first 50 cycles

Fig. 20

The total number of periodic data changes increases from 0 to 18; the average delay of highest priority packets

Fig. 21

The total number of periodic data changes increases from 0 to 18; the total delay

When the total number of periodic data changes increases from 0 to 3, the change in the volume of data is less than the maximum number of Tx-Beacons that can be transmitted during the initial wait time. Mn and Cn for the first 50 cycles are shown in Fig. 22. The average delays of high priority packets under the DWT-MAC and QPPD-MAC protocols are shown in Fig. 23. The average delay of high priority packets is greater for the former in this case. When the total number of ordinary nodes is 6, 12 or 18, the average value of highest priority packets increases by 4.7%, 6% or 4.6%. Thus, in this case, the average delay of high priority packets is slightly higher under the DWT-MAC protocol. The total delays of DWT-MAC and QPPD-MAC protocols are shown in Fig. 24. When the number of ordinary nodes in network is 6, the total delay for the DWT-MAC protocol is higher than that of the QPPD-MAC protocol, but when the number of ordinary nodes is 12 and 18, the total delay of the DWT-MAC protocol is less than that of the QPPD-MAC protocol. When the volume of data periodically changes from 0 to 3, as the number of ordinary nodes increases, data are more likely to be distributed to different ordinary nodes. The number of cycles with at least three sender nodes increases as well. This adjusts the wait time of the DWT-MAC protocol to T_w = 4T_tx more often, leading the total delay of the DWT-MAC protocol to be higher.

Fig. 22

The total number of periodic data changes increases from 0 to 3; Mn and Cn in the first 50 cycles

Fig. 23

The total number of periodic data changes increases from 0 to 3; the average delay of highest priority packets

Fig. 24

The total number of periodic data changes increases from 0 to 3; the total delay

Thus, the total volume of periodic data changes by a value less than the maximum number of Tx-Beacons that can be transmitted in the initial wait period. The average delay of high priority packets under the DWT-MAC protocol is less than that of the QPPD-MAC protocol. As this change becomes more significant and as the total number of ordinary nodes in the network increases, the average delay of high priority packets decreases more under the DWT-MAC protocol, but the total delay of the DWT-MAC protocol is greater than that of the QPPD-MAC protocol. When the total volume of data in each cycle changes periodically by less than the maximum number of Tx-Beacons that can be transmitted in the initial wait period, the delay corresponds to different numbers of ordinary nodes. Whether the delay of the DWT-MAC protocol is less than that of the QPPD-MAC protocol depends on the total number of ordinary nodes. As the total number of ordinary nodes increases, the delay of the DWT-MAC protocol is more likely to become higher than that of the QPPD-MAC protocol.

The total volume of data is constant

This section illustrates delays that occur when the total volume of data remains the same in each cycle. DWT-MAC and QPPD-MAC protocol delays are evaluated under different conditions.

First, the total volume of data for each cycle is equal to 3, and Mn and Cn for the first 50 cycles are shown in Fig. 25. In this case, average delays of high priority packets for the DWT-MAC and QPPD-MAC protocols are shown in Fig. 26. Regardless of the total number of ordinary nodes, the average delay of high priority packets under the DWT-MAC protocol is equal to that of the QPPD-MAC protocol. The total delays of the DWT-MAC and QPPD-MAC protocols are shown in Fig. 27. The total delay of the DWT-MAC protocol is higher than that of the QPPD-MAC protocol, as the wait time of the DWT-MAC protocol dynamically varies between 2 T_tx, 3 T_tx, and 4 T_tx in this case, causing the total delay of the DWT-MAC protocol to increase. As the total number of ordinary nodes increases, the total delay of the DWT-MAC protocol increases more, and as the total number of ordinary nodes increases, new data are more likely to be distributed to different ordinary nodes. The number of cycles with at least three nodes participating in transmission increases as well.

Fig. 25

The total volume of data is equal to 3 in each cycle; Mn and Cn in the first 50 cycles

Fig. 26

The total volume of data is equal to 3 in each cycle; the average delay of high priority packets

Fig. 27

The total volume of data is equal to 3 in each cycle; the total delay

We increase the total volume of data in each cycle to 6. Mn and Cn for the first 50 cycles are shown in Fig. 28. The average delays of high priority packets of the DWT-MAC and QPPD-MAC protocols are shown in Fig. 29. When the volume of data in each cycle is greater than 3, the average delay of high priority packets of the DWT-MAC protocol is less than that of the QPPD-MAC protocol. When the total number of ordinary nodes is 6, 12 or 18, the average delay of high priority packets of the DWT-MAC protocol respectively decreases by 20.6%, 37.7% or 43.1% relative to the QPPD-MAC protocol. The average delay of high priority packets of the DWT-MAC is decreased significantly in this case. The total delays of the DWT-MAC and QPPD-MAC protocols are shown in Fig. 30, and that of the DWT-MAC protocol is higher because its wait times are always longer to transmit high priority data first, thus increasing delays.

Fig. 28

The total volume of data is equal to 6 in each cycle; Mn and Cn in the first 50 cycles

Fig. 29

The total volume of data is equal to 6 in each cycle; the average delay of high priority packets

Fig. 30

The total volume of data is equal to 6 in each cycle; the total delay

We set the total volume of data for each cycle to less than 3 and equal to 2, and Mn and Cn for the first 50 cycles are shown in Fig. 31. The average delays of high priority packets of the DWT-MAC and QPPD-MAC protocols are shown in Fig. 32 and they are equal in this case as observed when the total volume of data is equal in each cycle. The total delays of the DWT-MAC and QPPD-MAC protocols in this case are shown in Fig. 33, and that of the DWT-MAC protocol is lower since under the QPPD-MAC protocol, each cycle has two sender nodes with data to transmit, wasting time waiting, while the DWT-MAC protocol can adjust the wait time to 2T_tx, leading to a lower average delay of high priority packets and lower total delays.

Fig. 31

The total volume of data is equal to 2 in each cycle; Mn and Cn in the first 50 cycles

Fig. 32

The total volume of data is equal to 2 in each cycle; the average delay of high priority packets

Fig. 33

The total volume of data is equal to 2 in each cycle; the total delay

Therefore, when the total volume of data per cycle remains the same and is equal to the maximum number of Tx-Beacons that can be transmitted in the initial wait period, the average delay of high priority packets under the DWT-MAC protocol is equal to that of the QPPD-MAC protocol, and the total delay of the DWT-MAC protocol is greater than that of the QPPD-MAC protocol. When the total volume of data is less than the maximum number of Tx-Beacons that can be transmitted in the initial wait period, the average delay of high priority packets under the DWT-MAC protocol is still equal to that of the QPPD-MAC protocol, but the total delay of the DWT-MAC protocol is less than that of the QPPD-MAC protocol.

When the volume of data is greater than the maximum number of Tx-Beacons that can be transmitted in the initial wait period, the average delay of high priority packets under the DWT-MAC protocol is less than that under the QPPD-MAC protocol, but the total delay of the DWT-MAC protocol is higher than that of the QPPD-MAC protocol.

The total number of data irregularly changes

This section illustrates the delay that occurs when the total volume of data randomly changes. The delays of the DWT-MAC and QPPD-MAC protocols are evaluated under different conditions.

The total volume of data randomly generated is increased from 0 to 6. Mn and Cn in the first 50 cycles are shown in Fig. 34. The average delays of high priority packets under the DWT-MAC and QPPD-MAC protocols are shown in Fig. 35; when the total number of ordinary nodes is 6, 12 or 18, the average delay of high priority packets under the DWT-MAC protocol is decreased by 3.2%, 10.8% or 11%, respectively, relative to the QPPD-MAC protocol. The total delays of the DWT-MAC and QPPD-MAC protocols are shown in Fig. 36, and that of the DWT-MAC protocol is higher in this case.

Fig. 34

The total volume of data randomly changes from 0 to 6; Mn and Cn inthe first 50 cycles

Fig. 35

The total volume of data randomly changes from 0 to 6; the average delay of high priority packets

Fig. 36

The total volume of data randomly changes from 0 to 6; the total delay

We increase the scope of total data volume change with volumes randomly increasing from 0 to 12. Mn and Cn in the first 50 cycles are shown in Fig. 37. The average delays of high priority packets for the DWT-MAC and QPPD-MAC protocols are shown in Fig. 38. In this case, the average delay of high priority packets under the DWT-MAC protocol is further reduced, and when the total number of ordinary nodes is 6, 12 or 18, delays respectively decrease by 22%, 37.6% or 38% relative to the QPPD-MAC protocol. The total delays of the DWT-MAC and QPPD-MAC protocols are shown in Fig. 39, and that of the DWT-MAC protocol is higher in this case.

Fig. 37

The total volume of data randomly changes from 0 to 12; Mn and Cn in the first 50 cycles

Fig. 38

The total volume of data randomly changes from 0 to 12; the average delay of high priority packets

Fig. 39

The total volume of data randomly changes from 0 to 12; the total delay

We further increase the scope of data volume changes with volumes randomly increasing from 0 to 18. Mn and Cn in the first 50 cycles are shown in Fig. 40. The average delays of high priority packets for the DWT-MAC and QPPD-MAC protocols are shown in Fig. 41, and when the total number of ordinary nodes is 6, 12 or 18, delays respectively decrease by 30%, 45.6% and 47% under the DWT-MAC protocol relative to the QPPD-MAC protocol. Under the DWT-MAC protocol, the average delay of high priority packets decreases significantly in this case. Additionally, as the number of ordinary nodes increases, the average transmission delay of highest priority packets decreases more under the DWT-MAC protocol. The total delays of the DWT-MAC and QPPD-MAC protocols are shown in Fig. 42, and that of the former is still higher in this instance.

Fig. 40

The total volume of data randomly changes from 0 to 18; Mn and Cn in the first 50 cycles

Fig. 41

The total volume of data randomly changes from 0 to 18; the average delay of high priority packets

Fig. 42

The total volume of data randomly changes from 0 to 18; the total delay

We decrease the scope of data volume changes with volumes randomly increasing from 0 to 3. Mn and Cn in the first 50 cycles are shown in Fig. 43. The average delays of high priority packets for the DWT-MAC and QPPD-MAC protocols are shown in Fig. 44, when the volume of data is less than or equal to the maximun number of Tx-Beacons that can be transmitted in the initial wait period, the average delay of high priority packets under the DWT-MAC protocol is higher than that under the QPPD-MAC protocol, and when the total number of ordinary nodes is 6, 12 or 18, delalys respectively increase up 2.6%, 5.2% or 4.3% under the DWT-MAC protocol relative to the QPPD-MAC protocol. The total delay for this case is shown in Fig. 45. From the results, the total delay of the DWT-MAC protocol is decreased relative to the QPPD-MAC protocol.

Fig. 43

The total volume of data randomly changes from 0 to 3; Mn and Cn in the first 50 cycles

Fig. 44

The total volume of data randomly changes from 0 to 3; the average delay of high priority packets

Fig. 45

The total volume of data randomly changes from 0 to 3; the total delay

Experiment summary

The experimental results are summarized as follows: no matter the volume of data changes in what way, when the total volume of data changes by a value greater than the maximun number of Tx-Beacons that can be transmitted in the initial wait period, the average delay of high priority packets under the DWT-MAC protocol is less than that under the QPPD-MAC protocol, but the total delay of the DWT-MAC protocol is higher relative to the QPPD-MAC protocol, as the scope of data volume or the total number of ordinary nodes increase, the greater the proportion of the average delay of high priority packets and the total delay change under the DWT-MAC protocol relative to the QPPD-MAC protocol. When the total volume of data changes by a value less than or equal to the maximum number of Tx-Beacons that can be transmitted in the initial wait period, experimental results show that even when the average delay of high priority packets under the DWT-MAC protocol is increased relative to the QPPD-MAC protocol, the proportion of increase is not large. In most cases, when the average delay of high priority packets under the DWT-MAC protocol is decreased relative to the QPPD-MAC protocol, the total delay under the DWT-MAC protocol is increased relative to the QPPD-MAC protocol.

Conclusion

In the previous receiver-initiated MAC protocols for prioritized data, the wait time is seted as a fixed duration and thus leads to long delays in data transmission and high network energy consumption. The DWT-MAC protocol is proposed in this paper for reduceing the delay in data transmission as well as network energy consumption, the main innovation of the DWT-MAC protocol is that it is based on a novel dynamic timer adjustment algorithm, it can dynamically adjust the wait times in each cycle according to the number of Tx-Beacons received by receiver. The preformance of the DWT-MAC protocol is evaluated in simulations, the results show that by dynamically adjsting the wait time, when the total volume of data changes greatly in the network, the DWT-MAC protocol can significantly reduce the average end-to-end delay of high priority packets, and when the total volume of data changes not greatly in the network, the DWT-MAC protocol can also significantly reduce the total end-to-end delay.