Sage Journals: Discover world-class research

Abstract

Wireless Sensor Networks are a key element of the Internet of Things since they are one of the most interesting pervasive systems. However, they are battery-powered, so energy efficiency represents one of the primary design challenges to address. This goal can be primarily achieved through an optimization of the communication procedure, which is the most power-consuming component of a WSN node. Nevertheless, the poor resources of the embedded devices usually limit the complexity of protocol solutions, which thus are not sufficient to reach satisfactory results. For this reason, the current trend aims at integrating both hardware and software solutions. In this work, a cross-layer solution, based on the combined use of a duty-cycling MAC protocol and a reconfigurable beam-steering antenna, is presented and validated. It significantly reduces the nodes' power consumption by exploiting scheduling techniques and directional communications. Specifically, a MAC scheduler manages the activation of the antenna sectors based on information coming from both MAC and network layers. This way, node awakenings occur only when an actual communication has to take place and only the interested antenna sector is activated. The effectiveness of the proposed cross-layer approach has been evaluated through OMNeT++ simulation tool.

1. Introduction

The modern Internet is strongly oriented towards the Internet of Things (IoT), that is, a new vision of Internet according to which everyday objects are becoming proactive actors of the global Internet, with the capability of generating and consuming information for advanced applications [1]. One of the most flexible technologies for this new vision is the wireless technology, because the absence of wired connections makes physical devices more useful in a widespread set of real scenarios. In this perspective, Wireless Sensor Networks (WSNs) are the ideal choice since sensor nodes are able to self-configure and self-organize. These characteristics allow deploying WSNs even in hostile environments in order to capture parameters (e.g., temperature, light, and humidity) without human intervention. Then, exploiting the wireless channel and the multihop communication among nodes, the collected data are sent to a central processing point or are exploited by user-customized mash-up applications [2]. Other strengths of this technology are represented by the low cost of devices, their small size, and their low power consumption. These simple yet fundamental functionalities are of great interest for a plethora of applications, such as building automation, surveillance, military operations, healthcare, and logistics, just to mention a few of them. However, to properly work, a WSN has to never be partitioned; that is, a path to the sink should always be present. With regard to this aspect, it is important to note that nodes are battery-powered and, in most cases, such a power cannot be regenerated or replaced; therefore its exhaustion corresponds to a definitive shutdown of the node. In this situation, not only data collected by that node but also some network routes are lost. Of course, a large number of node losses, due to battery depletion, could lead to network partitioning. So, the energy consumption is a key aspect in WSNs and it is crucial that sensor nodes optimize power consumption to extend the network lifetime in a consistent way with real use cases.

Let us observe that the power consumption of nodes is negligible during data sensing and processing procedures. On the contrary, the data communication towards the central processing system has a strong impact on the nodes' battery. This issue has a twofold cause: on the one hand, the radio transceiver is the most power-consuming component of a WSN node by itself; on the other hand, the communication phase is associated with phenomena such as collisions, overhearing (i.e., listening of messages addressed to another node), overemitting (i.e., transmission of data to a node that cannot receive them), and idle listening (i.e., listening to the channel in absence of communications), which substantially reduce the nodes battery. For these reasons, many works in literature are focused on energy saving, with particular emphasis on MAC layer, since it is responsible for managing channel access control mechanisms. In particular, many MAC-based energy saving solutions exploit the possibility to tune the duty-cycle, so enabling nodes to switch their radio between ON and OFF state according to a predefined scheduling. Unfortunately, protocol solutions cannot be too complex to not negatively affect the limited resources of the embedded devices. These issues often suggest combining new MAC protocols with power-saving hardware solutions. Among these, the use of directional or switched-beam antennas is one of the most explored. Traditionally, WSN nodes are equipped with an omnidirectional antenna, so most of the transmitted power is wasted towards useless directions. On the contrary, exploiting switched-beam antennas, it is possible to focus the transmission power only along the proper direction, so minimizing the power consumption. Nevertheless, the integration of beam-steering antennas with energy-efficient MAC protocols for WSNs has not been exhaustively explored, so it is still open to new solutions.

In this work, an energy-efficient cross-layer MAC protocol, called Hybrid Energy-Aware Cross-Layer MAC (HEC-MAC), based on the integrated use of a scheduling schema and a switched-beam antenna, is presented and validated. The proposed solution is able to reduce power consumption by avoiding transmissions towards unnecessary directions and by adapting the wake-up time of the radio transceiver to the network traffic. More in detail, during the setup phase, each node communicates to its neighbors the time interval during which it will transmit and, in addition, it detects all antenna sectors through which it can exchange data with its neighbors. Moreover, exploiting a cross-layer communication between network and MAC layers, each node is able to notify its neighbors about the identity of its next hop. In this way, during the steady state, a node wakes up only when its “previous hop” is transmitting and it turns on only the proper antenna sector. Finally, to reduce even more the power consumption, each node can anticipate the end of the wake-up period if no more packets have to be transmitted or received. The proposed solution is also able to face topology changes: every node updates its neighbor list as soon as it is not able to listen to transmission of a neighbor or it detects the presence of a new node.

In order to evaluate the effectiveness of the proposed cross-layer solution, it has been compared with three other protocols, which are described more in detail in the next section: AS2-MAC [3], AntDirMAC [4], and ZigBee [5]. The first one is a duty-cycle-based protocol that exploits a MAC scheduler to minimize the power consumption, the second one tries to maximize the network lifetime through the use of switched-beam antennas, and the third one is the protocol used by ZigBee that does not provide any specific energy-efficient mechanism.

The rest of the paper is organized as follows. Section 2 summarizes the state-of-the-art of energy-efficient solutions at MAC layer. The design of the MAC scheduler is described in Section 3. The simulation model is presented in Section 4, while in Section 5 numerical results are discussed. Conclusions are drawn in Section 6.

2. Related Works

The duty-cycle-based MAC protocols for WSNs fit into three main categories: preamble-sampling, scheduling, and hybrid approaches.

Preamble-sampling MAC protocols exploit the technique of Low Power Listening (LPL) [6] in order to sample the preamble of the packets. When there are no packets to be exchanged, LPL minimizes the duty-cycle, but it needs a preamble longer than the wake-up period to assure that the receiver can detect the channel activity. Examples of protocols belonging to this category are BMAC [7], WiseMAC [8], X-MAC [9], and XY-MAC [10]. Among these, WiseMAC and X-MAC are the most advanced. Through WiseMAC, which combines nonpersistent Carrier Sense Multiple Access (CSMA) with preamble sampling to mitigate idle listening, node schedulers are independent and information about the next awake period is piggybacked in the data acknowledgement frame. This value is used to dynamically determine the next awake time, allowing node to use short preamble. On the other hand, X-MAC is a particular preamble-sampling protocol that sends a strobed preamble, that is, a sequence of short preambles each of which is followed by a pause. A preamble packet contains the address of the target node, so that the destination node can recognize its own IP address as soon as it receives the preamble and can immediately reply with an ACK (early acknowledgment) during the short free interval. When the sender receives an ACK packet, it stops sending preamble. A recent enhancement of X-MAC is XY-MAC, which proposes a technique, called early termination, to sharpen the size of X-MAC pauses. With sharpened pauses, the sampling period at receivers can be effectively reduced to acquire a high level of energy efficiency, especially under low traffic loads.

MAC protocols based on scheduling approach manage duty-cycling by periodic synchronization messages and packet transmissions. One of the first duty-cycle MAC protocols for WSNs is S-MAC [11], according to which all nodes in a neighborhood wake up simultaneously and listen to the channel. They remain awake during the entire awake period even if they are neither sending nor receiving data, so leading to high latency and low throughput. Instead, RMAC [12] exploits cross-layer routing information in order to avoid latency. In particular, a setup control frame is used to schedule the upcoming data packet delivery along a specific route, so that an upstream node can send the data packet to intermediate relay nodes, which can immediately forward it to the downstream node. Recent works on scheduling-based protocols are [13–15]. In particular, [13] presents a TDMA-based approach. In [13], authors propose a MAC protocol, called LASA, in which time slot is not static but has a dimension that varies dynamically depending on the traffic load in the network. In order to achieve this result, the network is first organized into clusters and, in each of them, a cluster head is elected for assigning time slots to all nodes in the cluster. In [14], authors propose eL-MAC protocol that reduces idle listening, overhearing, and hidden terminal problems and in addition decreases the power consumption by allowing selecting different duty-cycle depending on the current data rate. Each node uses a beacon message to notify the destination node about the presence of data to be received, allowing nodes to switch to the sleep mode if there are no transmissions. Finally, in [15], a cross-layer approach is represented by CL-MAC that allows obtaining significant energy savings by using network-layer information. In particular, this protocol seeks to maintain the nodes in a sleep state as long as possible, by modifying RTS and CTS packets to inform each node whether it is included in the routing path of the current transmission or not.

Hybrid solutions combine some features of preamble sampling with scheduling techniques. A meaningful example is AS-MAC [16], which is used to asynchronously coordinate the wake-up times of neighboring nodes to reduce overhearing, contention, and delay. Furthermore, it exploits LPL to minimize the periodic wake-up time. Since nodes store the wake-up schedules of their neighbors, they know when these turn active. One of the main disadvantages of AS-MAC is the inefficiency in packet broadcasting, since it has to transmit every broadcast packet once per each neighbor. A protocol that overcomes the weaknesses of AS-MAC is AS2-MAC [3], which is based on an efficient setting of the node's duty-cycle as a function of the transmission times of the neighbor nodes. In a duty-cycle period, each node wakes up once to transmit and N times to receive, where N is the number of neighbors, while it remains in sleep mode for the rest of the time.

In order to improve the performance of MAC solutions without burdening the limited resources of WNS nodes, several works in the literature exploit the combined use of software algorithms and hardware solutions. In [17], a protocol that uses a directional antenna and a busy tone to ensure the energy saving is proposed. It exploits the mechanism of RTS/CTS to discover the location of its neighbors and it stores this information in a cache table. Messages that need to be sent in broadcast force the antenna to work in omnidirectional mode, whereas unicast messages are sent in one direction using the information stored in the cache table. Instead, the Direction Antenna at Sink (DAaS) protocol [18] extends the lifetime of the network by increasing the transmission range of the sink and by scheduling wake-up and sleep times for nodes according to the SMAC protocol. In particular, by changing the transmission range of the sink, this protocol distributes more fairly the power consumption among nodes communicating directly with the sink. In [19], the authors propose the Mobile Synchronous Transmission Asynchronous Reception Directive (MD-STAR) protocol, which is inspired to the WiseMAC and SMAC protocols, with the addition of directional antennas. In particular, the protocol achieves time-space synchronization in presence of mobile nodes and allows the management of the smart antenna adapting the Radio Frequency parameters in accordance with the characteristics of the communication link. In [20], authors present a MAC protocol, called Directional Utra-Wideband MAC (DU-MAC), which tries to reduce the power consumption by creating “omnidirectional” links between network nodes; that is, each directional link has a transmission range confined within the coverage area of an omnidirectional antenna. This solution results in a decrease of radiated power of about N times compared to a transmission in omnidirectional mode, where N is the number of antenna sectors. Finally, in [4], the AntDirMAC protocol is presented. It significantly reduces the nodes' power consumption by exploiting AS2-MAC protocol and directional communications. On the one hand, the scheduling algorithm of AS2-MAC reduces the number of collisions in the network and faces phenomena such as the hidden node problem, the idle listening, and the overhearing. On the other hand, the switched-beam antenna further reduces the power consumption by focusing communications only towards the proper sector.

3. The Proposed MAC Scheduler: HEC-MAC

The basic idea of the defined protocol is to ensure smart awakenings; that is, nodes wake up only when they actually have data to send or receive and only for the needed time. Moreover, by exploiting the switched-beam antenna, the transmission power is focused only on the desired direction in order to keep the power consumption as low as possible. More in detail, the proposed MAC schema consists of two main parts: the network start-up and the steady state. The network setup procedure, in turn, is characterized by three phases: (i) transmission time announcement, (ii) sectors discovery, and (iii) next hop announcement. During transmission time announcement phase, each node chooses its transmission time (i.e., the time instant at which it periodically can transmit) and then communicates such information to its neighbors. During the first phase, every node detects where its neighbors are located, that is, the antenna sector useful to reach each neighbor. Finally, during next hop announcement, every node notifies the network about the identity of its next hop. During the steady state, the information gathered through network start-up is used to optimize both outcoming and incoming transmissions. Indeed, for each duty-cycle period, a node wakes up at most one time for transmission (if it has data to transmit) and N times for receiving, where N is the number of “previous hop” nodes (if these nodes have data to transmit). Of course, if a transmission occurs, only the needed antenna sectors are exploited.

For the sake of clearness, before describing the new scheduler in detail, some parameters used in the discussion are introduced below. (i)

$T_{0}$ is the time interval (in seconds) between two subsequent transmissions. It is the same for every node and it is preconfigured (Figure 1).

(ii)

WakeTime is the maximum time interval (in seconds) for transmitting local data or for receiving packets from neighboring nodes (Figure 1).

(iii)

SleepTime is the minimum time interval (in seconds) that each node spends in sleep mode (Figure 1).

(iv)

Announce Packet ( $Pk t_{ANN}$ ) is a signaling packet used by each node to advertise its presence; it contains the time interval between the current time and the next awakening time chosen for transmission.

(v)

Alert Packet ( $Pk t_{ALERT}$ ) is a signaling packet used by a node to alert a neighbor about a possible transmission overlapping.

(vi)

Full Packet ( $Pk t_{FULL}$ ) is a signaling packet used by a node to inform its neighbors that it is out of the network.

(vii)

Reservation Packet ( $Pk t_{RSV}$ ) is the packet used by a node to reserve the channel for the sector discovery phase.

(viii)

Discovery Packet ( $Pk t_{DSC}$ ) is a unicast packet used to identify which neighbors are reachable through each antenna sector.

(ix)

NextHopIndication Packet ( $Pk t_{NEXT}$ ) is a packet used by a node to inform its neighbors about the identity of the next hop.

(x)

Wake-Up Table ( $W_{TBL}$ ) is a data structure used by WSN nodes to store information about its neighbors. Each table entry contains the following information: (a) the ID of the neighbor, (b) the offset of its awakening time, (c) the antenna sector for communicating with the neighbor, and (d) a Boolean value indicating whether the neighbor is a “previous hop” or not.

In the following, both the network start-up phase and the steady state are described.

Figure 1

Duty-cycle parameters.

3.1. Network Start-Up

As said in the previous section, the first step of the network start-up consists in the exchange of information about the transmission time. To do so, WSN nodes send $Pk t_{ANN}$ s to their neighbors. On the reception of such a message from an unknown neighbor, the HEC-MAC protocol updates its $W_{TBL}$ by storing a new entry. However, before being stored, the information on the transmission time of the neighbor must be validated; that is, the time chosen by the new neighbor has not to overlap with the transmission intervals of the other neighboring nodes already stored into its $W_{TBL}$ . If the validation succeeds, the transmission time of the neighbor is converted into offset to be consistent with the time view of the receiving node and then it is stored with the other offsets in ascending order. Otherwise, if some overlaps occur, the receiving node sends a $Pk t_{ALERT}$ to its neighbor, specifying the overlapping interval. In such a case, the neighbor stores the received information into its $W_{TBL}$ and chooses a new transmission time. This mechanism also reduces one of the main problems that afflict ad hoc networks, that is, the hidden node problem: by leveraging the $Pk t_{ALERT}$ , collisions among nodes two hops away are avoided. If the new node cannot find a valid transmission time, that is, the network is full, it broadcasts a $Pk t_{FULL}$ and turns off the radio. On the reception of such a message, all the neighbors delete the entry eventually stored for that node. It is worth noting that the probability of a missed join can be controlled through an accurate tuning of network parameters, first of all the duty-cycle period ( $T_{0}$ ) and the WakeTime interval. Indeed, if the duty-cycle period is increased and the WakeTime is decreased, then a new node has more chances to find a free time slot for transmission (and, in addition, all the nodes save more energy). However, under a heavy traffic load, this setup's conditions could lead to a rapid worsening of the network performance due to an increase of the buffer overflow events (i.e., during the sleep interval, nodes' buffers have high probability to be filled of packets exceeding their maximum size). The optimal combination of network parameters, as well as the tradeoff between “joining probability/energy saving” and “packet delivery,” should be evaluated case by case. Figure 2 shows a flow chart that describes this first step of the network setup.

Figure 2

Flow chart of the transmission time announcement phase.

Let us now describe how a node chooses its own transmission time. As said, the timeline is divided into fixed slot of $T_{0}$ seconds and each node has to choose its own transmission time as a random value in this interval, also taking into account the choices done by its neighbors. This separation in time among transmissions of neighboring nodes leads to reduce the contention for the channel access. More in detail, if the $W_{TBL}$ is empty, then the transmission time is randomly selected in the interval

\begin{matrix} [0, T_{0} - (WakeTime + 2 \cdot TurnAroundTime)], \end{matrix}

(1)

where WakeTime is the time window dedicated to data transmission and TurnAroundTime is the amount of time the radio needs for changing its state. If the

W_{TBL}

is not empty, then the node tries to set its own transmission time to a value different from those of its neighbors, in order to avoid collisions due to simultaneous transmissions. In particular, the node checks if there are two consecutive entries in the table, namely, ith and

(i + 1)

th, whose offsets difference is greater than

\begin{matrix} 2 \cdot WakeTime + 4 \cdot TurnAroundTime . \end{matrix}

(2)

If so, the transmission time is chosen within the interval

\begin{matrix} [offset [i] + D, offset [i + 1] - D], \end{matrix}

(3)

where

D = WakeTime + 2 \cdot TurnAroundTime

, whereas

offset [i]

and

offset [i + 1]

are the offsets associated with the ith and

(i + 1)

th entries, respectively. Note that the node also checks the time intervals:

\begin{matrix} [0, offset [0]], \\ [offset [n], T_{0} - D], \end{matrix}

(4)

where

offset [0]

and

offset [n]

are the offsets associated with the first and last entry, respectively. In order to maximize the probability that all its neighbors receive the message, a node sends the

Pk t_{ANN}

three times.

During the sector discovery phase, each node learns where the neighbors previously stored in $W_{TBL}$ are located; that is, it detects the proper antenna sector for communicating with each neighbor. This mechanism allows obtaining a reduction of power consumption during both transmission and reception phase. Indeed, in both cases, the node enables only the antenna sector useful to communicate with the desired neighbor. A node performs the localization phase by broadcasting, during its transmission time, a Pkt_RSV simultaneously in all directions. Through this packet, it communicates the total amount of time needed for the discovery phase, which is equal to the number of neighbors multiplied by the number of sectors and by the time interval required for sending a packet and receiving the corresponding acknowledgment. After the channel reservation, a node sends a unicast Pkt_DSC on each antenna sector and waits a response for a short time period. The node that receives a location request via a Pkt_DSC responds with the same kind of packet, also including the Received Signal Strength Indicator (RSSI) value related to the received request. In fact, since the radiation lobes of adjacent antenna sectors may overlap, the discovering node may receive more responses (relating to different sectors) from the same neighbor; in this case, it stores the sector corresponding to the response with the highest RSSI value. Figure 3 summarizes the sector discovery procedure.

Figure 3

Flow chart of the sector discovery phase.

The last step of the network start-up is the next hop announcement, which is carried out after the convergence of the routing algorithm. This mechanism is very useful because, except for broadcast communications, a node sends packets only to its next hop and receives packets only from its “previous hop” nodes. So, knowing in advance the previous hops allows keeping OFF the radio transceiver if the incoming communications have not to be received. To carry out this mechanism, upon the convergence of the routing protocol, the network layer of each node communicates to the MAC layer the identity of the next hop. This information is then broadcasted, through a Pkt_NEXT, to all neighbors, which store it in their $W_{TBL}$ . As explained in the next section, this result allows a node to immediately sleep if the incoming communication is not initiated by a “previous hop” node. Of course, the next hope announcement is triggered every time the routing algorithm is fired due to topology changes.

3.2. Steady State

In this phase, each node knows (i) the transmission times and (ii) sectors of its neighbors, as well as (iii) the identity of its previous hops. In the steady state, two kinds of periodic events, namely, the transmission and the reception of a packet, may happen. With regard to the periodic events, the node sets a timer for the next scheduled event in its $W_{TBL}$ . When the timer expires, if the scheduled event is data transmission, the node checks the presence of packets to be transmitted in its queue and eventually activates the proper antenna sector and starts the transmission for WakeTime seconds. For each transmitted packet, the node waits for an ACK from the intended receiver. If no ACK is received, the packet is sent again. To adapt the WakeTime to the data rate in order to avoid useless idle periods, the node early terminates its transmission session (i.e., before the end of WakeTime time interval) if no more packets have to be sent. Moreover, also the receiving node can take advantage of this approach since the early termination is notified in the last packet transmitted by the sender. This feature gives to each WSN node a kind of traffic-awareness capability and allows significantly reducing the power consumption when the traffic load is low, which is a very common situation in several scenarios for WSNs. At the end of its transmission interval, the node schedules the next event of $W_{TBL}$ . When a receiving event is scheduled, the node starts listening to the channel exploiting the antenna sector related to the transmitting neighbor. If a packet is received, it sends an ACK. On the contrary, if nothing is received the node immediately turns off its radio transceiver. A particular transmission/reception event is represented by broadcast communications, which would require a number of communication sessions equal to the number of antenna sectors. To avoid this useless approach, the sending node sends eventual broadcast packets at the beginning of its transmission interval by turning on all the antenna sectors. On the other hand, its neighbors activate the proper antenna sector for a short time interval to eventually detect broadcast communications. If no transmissions are detected, the neighbors turn off their radio component and only the next hop of the sending node reactivates the radio ON to receive the unicast data. Finally, it is important to note that inaccurate synchronization may occur among the various node timers, potentially leading to idle listening and overemitting phenomena, in addition to possible data losses. In order to address this issue, a receiver anticipates the wake-up for reception by a quantity of $I_{G} = 2 \cdot C_{DRIFT}$ with respect to the scheduled time, where $C_{DRIFT}$ is the maximum clock drift that may occur (Figure 4). Figure 5 summarizes the main scenarios that may occur during the steady state. During the first duty-cycle period, the transmitting node (Node 1) has nothing to send, so its neighbors activate the proper antenna sector only for a very short time interval before being aware that no broadcast communications are coming. Then, the next hop node (Node 2) reactivates the radio transceiver waiting for a unicast transmission, but it immediately switches in sleep mode because no packets are received. During the second duty-cycle period, Node 1 does not broadcast any packets but sends some data to its next hop. Since the amount of these data does not need all the WakeTime to be transmitted, an early termination event occurs, so both Node 1 and Node 2 anticipate the switch-off of their radio component. To better clarify how the radio transceiver changes its state according to the previously described behavior, Figure 6 shows the state diagram of the PHY layer and highlights the main actions that trigger a state change. It is worth noting that the CCA state in Figure 6 stands for Clear Channel Assessment, that is, the carrier sensing procedure performed before accessing the channel.

Figure 4

Early awakening to address clock drift.

Figure 5

Steady state behavior: minimization of idle periods of receiving nodes (Nodes 2 and 3) when the transmitting node (Node 1) has nothing to transmit or needs lesser time than WakeTime.

Figure 6

PHY layer state diagram.

4. Simulation Model

An exhaustive simulation campaign was carried out in order to appreciate the effectiveness and efficiency of the proposed MAC scheduler. The tool used to obtain this performance validation is the discrete event simulator Omnet++ [21]. The performance of the designed protocol was compared with the MAC protocol currently used in ZigBee-based WSNs [5] and with two energy-efficient protocols already presented in the literature, namely, AS2-MAC and AntDirMAC. While the MAC protocol of ZigBee does not provide any specific energy-efficient mechanism, AS2-MAC exploits a duty-cycle-based MAC scheduler to minimize the power consumption, and AntDirMAC maximizes the network lifetime through the use of switched-beam antennas. Additionally, assuming that each real solution should enable an easy and fast deploy of a WSN, all the nodes in the network are assumed to be peer and, for ZigBee, beacon-less networks are simulated.

The main objective of the protocol evaluation is to show how the scheduler provides significant energy savings without compromising the efficiency of data delivery. Consequently, the power consumption and the Packet Delivery Ratio (PDR) have been chosen as performance metrics. In particular, the evaluated power consumption represents the instantaneous power used only by the radio transceiver. In more detail, to calculate it, the amount of time that a node spends in reception, transmission, and idle states during a duty-cycle period is calculated. Then, each time interval is multiplied by the corresponding power consumption value (per unit of time) reported in Table 1. Adding the obtained values and dividing the result by the duration of the duty-cycle period ( $T_{0}$ ), the instantaneous power consumption of each node is found. On the other hand, the PDR is calculated as the ratio between the total number of packets that reach the sink and the sum of these packets with those lost in the network. In this way, the PDR is not affected by packets that are still on fly, so it is clearer how much the sleep period affects the PDR. These metrics are analyzed for different values of some important parameters such as the packet rate of nodes and the number of nodes in the network (i.e., the network density). By varying these parameters it is possible to evaluate the protocol behavior with respect to changes in interference, hop distance from the sink node, and traffic load.

Table 1

Simulation parameters.

Parameter	Value
Network topology	Squared grid
Application packet dimension [byte]	60
Buffer dimension [byte]	4000
Network bandwidth [kb/s]	250
Transmission power [dBm]	−10
Packet error rate	1%
Packet rate [pkt/s]	0.01, 0.02, 0.1
Number of nodes	16, 25, 36
WakeTime [s]	1
Duty-cycle period [s]	40
Number of antenna sectors	8
Instantaneous power in TX state	48.2 mW
Instantaneous power in RX state	48.2 mW
Instantaneous power in IDLE state	53.6 mW

The network layer of the considered model is compliant with the ZigBee specifics, whereas the MAC and PHY layers are implemented according to the IEEE 802.15.4 standard [22]. The proposed schema has been introduced into the MAC layer as a manager for the IEEE 802.15.4 MAC protocol.

The network topology is a squared grid topology and each node is provided with an on-board sensor that generates data packets with a fixed size of 60 bytes at the application layer. A Constant Packet Rate (CPR) traffic has been chosen, since WSNs are often used in applications aimed at periodic environmental monitoring. The WakeTime, that is, the time interval during which a node transmits or receives data packets, and the duty-cycle period $T_{0}$ have been set to 1 sec and 40 sec, respectively. It is important to highlight that the $T_{0}$ parameter, in combination with WakeTime, is crucial for both energy saving and network performance. If its value is increased (and/or WakeTime's value is decreased), then the WSN nodes save more energy, but they can seriously lose packets due to buffer overflows when the traffic load increases. For this reason, it is not possible to set an optimal value in general, but it should be evaluated for each specific scenario. We made this kind of analysis in [3]. For MAC solutions based on switched-beam antenna, eight sectors have been set. This choice has been done based on the analysis reported in [4], where it was demonstrated that eight sectors represent the best tradeoff between power saving and electromagnetic overlapping among antenna sectors. The frequency band used is 2.4 GHz and packets are discarded by the receiver with a packet error rate equal to 1%.

All the main simulation parameters are reported in Table 1. The simulation results are characterized by a 95% confidence interval with a 5% maximum relative error.

5. Results

The protocol solutions evaluated in this section are based on different design principles, so it is possible to deduce some preliminary considerations. Regarding ZigBee standard, (i) it does not provide any energy-efficient mechanism, (ii) the radio transceiver of each node is always active, and (iii) the network communications are carried out via omnidirectional transmissions. Moreover, negative phenomena such as idle listening and collisions are frequent due to the absence of any kind of synchronization. For these reasons, the power consumption of a node is expected to be high, leading to poor network lifetime. On the other hand, since the radio component is always ON, packets are sent as soon as they are available at the MAC layer, thus packet losses due to buffer overflow events should be unusual. This behavior should lead to a high PDR, especially in network characterized by low traffic. Instead, AS2-MAC is a low-power protocol that manages the duty-cycle of the node by exploiting an asynchronous MAC scheduler. So, it is expected to guarantee a significant reduction of power consumption. However, the awakenings of WSN nodes compliant with AS2-MAC are not so smart, since a node wakes up, for WakeTime seconds, every time a communication event (for both transmitting and receiving data) is scheduled. These awakenings occur also if no actual communications are coming and their duration is independent of the amount of transmitted data. Finally, AntDirMAC combines the pros of the scheduling approach with the advantages coming from directional communications. The use of switched-beam antennas should allow further minimizing the power consumption because lower power is necessary to cover the same distance during communication operations. Regarding the scheduling of node awakenings, AntDirMAC suffers from the same problems described for AS2-MAC. Finally, HEC-MAC makes nodes smarter since they are able to wake up only when an actual incoming transmission is occurring (i.e., a previous hop node is about to send data) and only for the needed time (by exploiting the early termination procedure). Of course, all the duty-cycle-based protocols could suffer from a lower PDR because buffer overflow events could happen during sleep periods.

The first performance analysis has been focused on evaluating the behavior of our solution when the network density (i.e., the number of nodes in the network) and the hop distance from sink node change. In this perspective, Figure 7 shows the trend of the power consumption for nodes one hop away from the sink, depending on the network density. The first consideration that arises from the graph concerns the increasing trend of the power consumption in relation to an increasing network density. However, this growth is much slower than one might imagine due to the next hop announcement mechanism, which allows WSN nodes to wake up only for receiving data coming from their “previous hop” nodes. In this way, although the network density increases, the number of “previous hop” nodes increases much more slowly. Furthermore, it is worth noting that the power consumption decreases when nodes send data at lower rate. This is because the early termination mechanism allows switching off the radio transceiver before the end of the WakeTime interval. Without this feature, the node would keep the idle state, which leads to about the same power consumption of transmissions. The trend of the power consumption in relation to the hop distance from the sink node is shown in Figure 8. In this case, the packet rate of each WSN node is equal to 0.1 pkt/s. According to the previous considerations, the graph shows a slow increase of power consumption when the network density increases. Moreover, it is interesting to observe the decrease of power consumption for nodes at the edge of the network. Indeed, they have less “previous hop” nodes with respect to nodes inside the network, so they can exploit the HEC-MAC features to keep their radio OFF for longer time.

Figure 7

Power consumption of HEC-MAC for nodes one hop away from the sink.

Figure 8

Power consumption of HEC-MAC for a packet rate of 0.1 pkt/s.

To highlight the effectiveness of HEC-MAC with respect to ZigBee, Figure 9 shows the comparison between the two protocols when a network with 36 nodes is considered. The packet rate of each node is equal to 0.1 pkt/s, and the power consumption is evaluated depending on the hop distance from the sink node. As expected, HEC-MAC provides a clear improvement on energy saving since it combines the advantages of both switched-beam antenna and duty-cycle approach. By exploiting directional transmissions, each node can reduce the power consumption by a factor approximately equal to the number of antenna sectors, whereas, thanks to the low duty-cycle, next hop information, and early termination, it wakes up for very short time interval in each duty-cycle period.

Figure 9

Comparison of the power consumption between HEC-MAC and ZigBee MAC, when there are 36 nodes and a packet rate of 0.1 pkt/s for each node.

A more meaningful evaluation about energy saving capabilities of HEC-MAC is highlighted in Figure 10, which shows a comparison among HEC-MAC, AS2-MAC, and AntDirMAC. As can be deduced from the graph, our solution leads to a significant reduction of power consumption. Indeed, as expected, the next hop information and the early termination procedure represent further features towards energy saving, since they allow WSN nodes to keep their radio OFF as much as possible. More in detail, AntDirMAC guarantees better performance with respect to AS2-MAC since it exploits directional communications, so reducing the power consumption of a factor equal to the number of antenna sectors. HEC-MAC further improves the energy management because it turns on the radio transceiver only for actual incoming transmissions. In this perspective, an interesting consideration can be made about the trend of the curves in the graph. While HEC-MAC shows a slow increase of the power consumption when the network density increases (as previously explained), both AS2-MAC and AntDirMAC reveal an increasing energy waste when there are more nodes in the network. This is because, when network density is higher, each node has more neighbors, so it has to wake up more times during a duty-cycle period (each time for WakeTime seconds), also if no actual packets are received.

Figure 10

Comparison of the power consumption among HEC-MAC, AS2-MAC, and AntDirMAC, when there are 36 nodes and a packet rate of 0.02 pkt/s for each node.

To evaluate the potential negative effect of HEC-MAC on network performance, the PDR has been evaluated by comparing the proposed solution with ZigBee standard, which guarantees the best performance regarding the packet delivery. The results of this comparison are shown in Figure 11, which highlights the trend of PDR in dependence of hop distance from the sink node. As expected, HEC-MAC provides lower PDR with respect to ZigBee due to the low duty-cycle. Indeed, between two successive transmissions, the data packets coming from upper layers are queued in the outgoing buffer. If the incoming rate of these packets is too high, several buffer overflow events may occur, so leading to packet losses and to a consequent decrease of PDR. However, since WSNs are usually exploited in loss-tolerant and low-rate applications, a little decrease in PDR is poorly significant. This is just the case of Figure 11, in which a packet rate consistent with a typical low-rate application for WSNs is chosen. In particular, the graph refers to a packet rate equal to 0.02 pkt/s and to a squared grid topology of 36 nodes. It is worth noting that the PDR decreases moving away from the sink node. Indeed, although the nodes at the edge of the network have to relay less packets, their own packets have to be relayed by several nodes, so the probability to lose those packets due to channel errors or buffer overflows is higher.

Figure 11

Comparison of the PDR between HEC-MAC and ZigBee MAC for nodes one hop away from the sink and with a packet rate of 0.1 pkt/s for each node.

6. Conclusions

The reduction of the power consumption is one of the major issues in WSNs, and the efficient management of the radio transceiver is the key element to achieve this objective. Actual trend is to combine software protocol solutions with hardware components, in order to reduce the energy waste without burdening the limited resources of WSN nodes. In this work, a cross-layer approach based on the joint use of a switched-beam antennas and a smart MAC scheduler is presented and validated. It significantly reduces the nodes' power consumption by exploiting scheduling techniques and directional communications. Specifically, the MAC scheduler manages the activation of the antenna sectors based on information coming from both MAC and network layers. Node awakenings occur only when an actual communication has to take place and only the interested antenna sector is activated. Moreover, each node is able to adapt its wake-up time to the network traffic, reducing it if lower data rate is required. The effectiveness of the proposed cross-layer approach has been evaluated through OMNeT++ simulation tool, which has been used to compare our solution with both ZigBee standard and two other energy-efficient MAC protocols, namely, AS2-MAC and AntDirMAC. The simulation results have shown that HEC-MAC protocol is able to further reduce the power consumption without negatively affecting the PDR in typical WSNs usage scenarios. We are aware that these results are only the first step towards the complete evaluation of our protocol solution, so we are already working in order to verify them through real testbeds and to perform further analysis in more complex scenarios.

Footnotes

Conflict of Interests

The authors declare that there is no conflict of interests regarding the publication of this paper.

References

Mainetti

Patrono

Vilei

Evolution of wireless sensor networks towards the Internet of Things: a survey

Proceedings of the 19th International Conference on Software, Telecommunications and Computer Networks (SoftCOM '11)

September 2011

16 21

2-s2.0-81455142290

Mainetti

Mighali

Patrono

Rametta

Discovery and mash-up of physical resources through a web of things architecture

Journal of Communications Software and Systems 2014 10 2 124 134

Anchora

Capone

Mighali

Patrono

Simone

A novel MAC scheduler to minimize the energy consumption in a wireless sensor network

Ad Hoc Networks 2014 16 88 104

10.1016/j.adhoc.2013.12.002

2-s2.0-84894457731

Catarinucci

Guglielmi

Mainetti

Mighali

Patrono

Stefanizzi

M. L.

Tarricone

An energy-efficient MAC scheduler based on a switched-beam antenna for wireless sensor networks

Journal of Communications Software and Systems 2013 9 2 117 127

2-s2.0-84886237081

ZigBee Alliance ZigBee Specification Document 053474r17

Merlin

C. J.

Heinzelman

W. B.

Duty cycle control for low-power-listening MAC protocols

IEEE Transactions on Mobile Computing 2010 9 11 1508 1521

10.1109/tmc.2010.116

2-s2.0-77956949428

Polastre

Hill

Culler

Versatile low power media access for wireless sensor networks

Proceedings of the 2nd ACM Conference on Embedded Networked Sensor Systems (SenSys '04)

November 2004

95 107

2-s2.0-26444451023

Enz

C. C.

El-Hoiydi

Decotignie

J.-D.

Peiris

WiseNET: an ultralow-power wireless sensor network solution

Computer 2004 37 8 62 70

10.1109/mc.2004.109

2-s2.0-4344668572

Buettner

Yee

G. V.

Anderson

Han

X-MAC: a short preamble MAC protocol for duty-cycled wireless sensor networks

Proceedings of the 4th International Conference on Embedded Networked Sensor Systems (SenSys '06)

November 2006

New York, NY, USA

ACM

307 320

10.1145/1182807.1182838

2-s2.0-34547480316

10.

Luo

Wang

XY-MAC: a short preamble MAC with sharpened pauses for wireless sensor networks

Proceedings of the IEEE Wireless Communications and Networking Conference (WCNC '12)

April 2012

1550 1554

10.1109/wcnc.2012.6214028

2-s2.0-84864334876

11.

Heidemann

Estrin

Medium access control with coordinated adaptive sleeping for wireless sensor networks

IEEE/ACM Transactions on Networking 2004 12 3 493 506

10.1109/tnet.2004.828953

2-s2.0-4344714071

12.

Saha

A. K.

Johnson

D. B.

RMAC: a routing-enhanced duty-cycle MAC protocol for wireless sensor networks

Proceedings of the 26th IEEE International Conference on Computer Communications (INFOCOM '07)

May 2007

Anchorage, Alaska, USA

1478 1486

10.1109/infcom.2007.174

2-s2.0-34548317545

13.

Hussain

Zahmati

A. S.

Fernando

LASA: low-energy adaptive slot allocation scheduling algorithm for wireless sensor networks

IEEE Sarnoff Symposium (SARNOFF '09)

April 2009

Princeton, NJ, USA

1 6

10.1109/sarnof.2009.4850360

2-s2.0-67650663279

14.

Latiff

L. A.

Rashid

R. A.

Syed Ariffin

S. H.

Wan Embong

W. M. A.

Fisal

Implementation of enhanced lightweight medium access (eL-MAC) protocol for wireless sensor network

Proceedings of the 16th Asia-Pacific Conference on Communications (APCC '10)

November 2010

Auckland, New Zealand

267 272

10.1109/apcc.2010.5679778

2-s2.0-79251563826

15.

Yang

Zhang

An energy efficient cross-layer MAC for wireless sensor networks

Proceedings of the International Conference on Electronic and Mechanical Engineering and Information Technology (EMEIT '11)

August 2011

Heilongjiang, China

683 686

10.1109/emeit.2011.6023141

2-s2.0-80053390484

16.

Jang

Lim

J. B.

Sichitiu

M. L.

AS-MAC: an asynchronous scheduled MAC protocol for wireless sensor networks

Proceedings of the 5th IEEE International Conference on Mobile Ad-Hoc and Sensor Systems (MASS '08)

October 2008

434 441

10.1109/mahss.2008.4660032

2-s2.0-67650652444

17.

Omar

E. A.

Elsayed

K. M. F.

Directional antenna with busy tone for capacity boosting and energy savings in wireless ad-hoc networks

Proceedings of the High-Capacity Optical Networks and Enabling Technologies (HONET '10)

December 2010

Cairo, Egypt

IEEE

91 95

10.1109/HONET.2010.5715751

18.

Cho

Lee

Kwon

Choi

Directional antenna at sink (DAaS) to prolong network lifetime in wireless sensor network

Proceedings of the 12th European Wireless Conference 2006—Enabling Technologies for Wireless Multimedia Communications (European Wireless '06)

April 2006

Athens, Greece

1 5

19.

Bencini

Collodi

Di Palma

Manes

An energy efficient cross layer solution based on smart antennas for wireless sensor network applications

Proceedings of the 4th International Conference on Sensor Technologies and Applications(SENSORCOMM '10)

July 2010

Venice, Italy

232 237

10.1109/sensorcomm.2010.43

2-s2.0-77957945309

20.

Karapistoli

Gragopoulos

Tsetsinas

Pavlidou

F.-N.

A directional MAC protocol with deafness avoidance for UWB wireless sensor networks

Proceedings of the IEEE International Conference on Communications (ICC '09)

June 2009

Dresden, Germany

1 5

10.1109/icc.2009.5198593

2-s2.0-70449500213

21.

OMNET++ Network Simulation Framework, October 2014, http://www.omnetpp.org/

22.

IEEE

LAN/MAN Standards Committee IEEE, Computer Society, IEEE standard for local and metropolitan area networks—part 15.4: low-rate wireless personal area networks (LR-WPANs)

IEEE Standard 2006 802.15.4-2006

IEEE

HEC-MAC: A Hybrid Energy-Aware Cross-Layer MAC Protocol for Wireless Sensor Networks

Abstract

1. Introduction

2. Related Works

3. The Proposed MAC Scheduler: HEC-MAC

3.1. Network Start-Up

3.2. Steady State

4. Simulation Model

5. Results

6. Conclusions

Footnotes

Conflict of Interests

References