Adaptive-opportunistic Aloha: A media access control protocol for unmanned aerial vehicle

Abstract

Recognizing that data collection in a unmanned aerial vehicle–wireless sensor network system is significantly different from that in a conventional wireless sensor network system, we propose a media access control protocol called adaptive-opportunistic Aloha for unmanned aerial vehicle–wireless sensor network systems. Based on a cross-layered design, this proposed adaptive-opportunistic Aloha protocol takes several important factors into consideration, including distribution of sensors, energy consumption, and transmission efficiency. In order for unmanned aerial vehicle to uniformly collect data from the ground sensors that are distributed in a random fashion, adaptive-opportunistic Aloha adopts a priority-based mechanism for channel assignment and collision avoidance, and importantly, the priority could be adaptively changed according to locations and distribution of the sensors. To improve energy efficiency, the adaptive-opportunistic Aloha can effectively put the sensors into the sleep mode when they do not need to send data, and they will be woken up by a beacon signal from the unmanned aerial vehicle for data transmission. Unlike opportunistic Aloha protocol, a well-known unmanned aerial vehicle–wireless sensor network media access control protocol, the adaptive-opportunistic Aloha adds a handshake into every time interval of transmission to help enhance the throughput with an acceptable level of system bit error rate. Experiment results have shown that the proposed media access control protocol can improve the overall throughput of unmanned aerial vehicle–wireless sensor network systems by more than 30% over opportunistic Aloha.

Keywords

Unmanned aerial vehicle wireless sensor network media access control protocol cross-layered design

Introduction

Over the last two decades, various types of wireless sensor network (WSN) systems have been proposed and successfully deployed in a slew of application areas spanning military, medical, environmental, and other businesses. Early WSN systems are merely static ones in the sense that all the sensor nodes are essentially stationary, and gradually, more WSN systems tend to include both stationary and mobile nodes for improved performance and/or added new functionalities. With continued technology breakthroughs in unmanned aerial vehicles (UAVs), there is a growing popularity that UAVs act as the mobile nodes in collaboration with the ground WSNs. Dubbed as UAV-WSN systems, they are able to perform jobs that are often considered difficult or even impossible for traditional WSNs.¹ For instance, the UAVs can fly over the area of interest (the ground or the sea surface), but hard to reach by humans or other means, to collect valuable sensor data. Another nice thing about the UAVs is that they can easily make multiple data collection trips within a short period of time when necessary.

The WSN-UAV systems have their own unique features and challenges. For instance, latency of data transmission between the sensors and the UAV is typically short, because they communicate with each other through direct links, as opposed to multi-hop communications seen in most conventional WSNs. One big challenge facing WSN-UAVs is that such systems must adopt a medium access control (MAC) protocol that guarantees high network throughput, low energy consumption, and small time intervals for sending data.²

There are various MAC layer protocols proposed for sensor networks. The Sensor MAC (S-MAC) protocol,³ for instance, exploits the idea of synchronization intervals that are managed at cluster level and periodic sleep–listen schedules are set in accordance with the synchronization. Neighboring nodes that form a virtual cluster can set up a common sleep schedule. If two neighboring nodes reside in two different virtual clusters, they wake up only at the listen periods of both clusters. As sensor nodes in S-MAC have to follow two different schedules, extra energy consumption caused by idle listening and overhearing is unfortunately unavoidable. Timeout-MAC (T-MAC)⁴ was later proposed to overcome the performance problem of the S-MAC under variable traffic loads. In T-MAC, synchronization during the listen periods within virtual clusters is broken, one of the main reasons causing the early sleeping problem. Similar to S-MAC and T-MAC, routing enhanced duty-cycle MAC (RMAC)⁵ also divides the operational cycle of a sensor node into three periods: SYNC, DATA, and SLEEP, but with a distinct difference. RMAC delivers a pioneer frame (PION) over multiple hops during the DATA period to set up a multi-hop schedule for subsequent data forwarding during the SLEEP period. As such, this approach can forward a data packet over several hops within one single cycle, thus cutting down the delivery latency. The RMAC protocol, however, does not consider data routing–related issues, which would cause significant performance degradation.

P-MAC⁶ considers cross-layer optimization with the goal of minimizing the communication overhead and maintaining the superiority of duty-cycling. It divides all the sensor nodes into different grades according to their logic hop distances to the sink. The lower a node’s grade, the fewer hops it needs to send its packets to a fixed data sink. As P-MAC, like S-MAC, T-MAC, and RMAC, does not consider changing network deployment situations, it is not a suitable scheme for a WSN system that includes a mobile data sink like a UAV.

As a sharp contrast to S-MAC, T-MAC, RMAC, and P-MAC, opportunistic Aloha (O-Aloha)⁸ is a scheme that incorporates channel state information into the Aloha protocol for sensor network with mobile agent.⁹ Each sensor estimates its fading state during the period when the mobile agent transmits a beacon. The node then flips a biased coin. If the outcome is “head,” the node transmits its packet. Otherwise, the node is silent for the current slot and restarts this process in the next slot. Although all the nodes use the same probability mass function, as they experience different fading conditions, the transmission probability of one node can be significantly different from that of others. There are lots of researches following O-Aloha scheme in the field of cross-layer design for WSN.^10–18 Furthermore, O-Aloha has been successfully adopted in other various wireless networks.^19–22

In Ho et al.,^7,23 a code division multiple access (CDMA)-based MAC protocol is dedicated for WSN-UAV systems. As this scheme only allows a limited number of active sensors to transmit their data at a given time, the total amount of time to complete data collection tends to be quite large, and there is no guarantee that each sensor will have an opportunity to send its data to the UAV even after all the transmission time windows have expired. Another big problem of this CDMA-based MAC and all the other aforementioned protocols is that they fail to deliver the uniformity of data collection. As a UAV needs to fly around to collect sensor data, and sensor nodes are typically not uniformly distributed on the ground, in the area where sensors are densely distributed, there is a high probability that data collision will happen, while in area where sensors are loosely distributed, UAV may not have sufficient data to collect. Either way, the UAV may have to make multiple flight trips to collect the right amount of data, causing a longer delay and higher power consumption.

To overcome the aforementioned problems, a new CDMA-based MAC protocol in the UAV-WSN systems is proposed in this article, where distribution density of sensors is considered to improve the system performance. Referred as adaptive-opportunistic Aloha (AO-Aloha), the proposed MAC protocol uses a Slot-Aloha like beaconing and adaptive priority–based channel assignment for channel allocation and collision avoidance. Each sensor node estimates its fading state during the period when the UAV transmits a beacon. In the beacon stage, head slots are assigned to the sensor nodes. During this period, a node that needs to communicate with the UAV transmits its head packet to the UAV for channel allocation and priority assignment. Nodes are partitioned in the mesh area by their locations. Priority of the node is assigned based on the node density of the area that the node belongs to. If a node in an area has a small number of nodes, this node will be assigned with a higher priority. If there is an unoccupied channel, the node that has the highest transmission priority will be allowed to claim the channel and then transmit its data packets to the UAV; all other nodes have to go back to silent mode for the remaining current slot and restart the process in the next slot. When two or more nodes have the same transmission priority, the transmission probabilities of these sensor nodes can still be different by taking into account the channel fading conditions.

The reminder of this article is organized as follows. Section “UAV-WSN system and models” introduces the UAV-WSN system as well as various models concerning the communications and energy consumption in such a system. The proposed AO-Aloha protocol is detailed in section “AO-Aloha: a CDMA-based MAC protocol for UAV-WSN systems,” followed by theoretical analysis on AO-Aloha’s bit error rate (BER) and throughput in section “Performance analysis.” Section “Simulation result” presents the simulation results of the proposed protocol. Finally, conclusions are drawn in section “Conclusion.”

UAV-WSN system and models

Communication model for UAV-WSN systems

Assume that wireless sensors are randomly deployed in the area, as depicted in Figure 1. Flying horizontally with constant speed of v at height of h, the UAV enters the sensor-covered area and continues to pass through along a predefined path, collecting the sensor data as it flies. The communications between the UAV and the ground sensors can be viewed as many to one multi-access communications, where the UAV acts as the mobile data sink; at any given time, the ground sensors that can directly communicate with the UAV form a circular region C with a radius of r, as shown in Figure 2.

Figure 1.

Sensors deployed on the ground and the UAV.

Figure 2.

The ground sensors within a circular region can communicate directly with the UAV. The sensors and the UAV form a communication cone, where the UAV is at the vertex, the base circle has a radius of R, and the cone has a height of h.

Assume that there are $N_{C}$ sensor nodes deployed over a circular area C which then can be divided into n equal sectors, and these sectors, respectively, have $N_{1}, N_{2}, \dots, N_{n}$ nodes. Let $S_{i}^{j}$ represent the state of the ith node at the jth sector in the circle. If node i successfully sends its data to the UAV, $S_{i}^{j} = 1$ ; otherwise, $S_{i}^{j} = 0$ . Accumulatively, the total number of sensors that have successfully sent their data to the UAV in the area C, denoted as $N_{S}$ , is given as

N_{S} = \sum_{j = 1}^{n} \sum_{i = 1}^{N_{j}} S_{i}^{j} = \sum_{i = 1}^{N_{1}} S_{i}^{1} + \sum_{i = 1}^{N_{2}} S_{i}^{2} + \dots + \sum_{i = 1}^{N_{n}} S_{i}^{n}

(1)

N_{C} = \sum_{j = 1}^{n} N_{j}

(2)

Typically, $N_{S} < N_{C}$ .

Packet error rate model for UAV-WSN systems

According to Venkitasubramaniam et al.,⁸ the system packet error rate (PER) of a UAV-WSN system, denoted as, is given as

PE R_{sys} = 1 - (1 - PE R_{mob}) (1 - PE R_{tran})

(3)

where $PE R_{mob}$ is the PER due to the movement of the UAV, and $PE R_{tran}$ is the PER caused by the packet transmission between a sensor and the UAV. More specifically, $PE R_{sys}$ is given as

\begin{array}{l} P E R_{s y s} = 1 - (1 - \frac{\sum_{i = 1}^{n} P E R_{m o b}^{P a r t i} * P k t_{i}}{N_{p k t}}) \\ (1 - \frac{\sum_{i = 1}^{n} P E R_{t r a n}^{P a r t i} * P k t_{i}}{N_{p k t}}) \end{array}

(4)

where N_pkt is the number of packets that are transmitted in the UAV-WSN system, ${PER}_{mob}^{Parti}$ is the PER due to the movement of the UAV above sector i, ${PER}_{tran}^{Parti}$ is the PER caused by the packet transmission between the sensor and the UAV at sector i, and $Pk t_{i}$ is the number of packets that are transmitted at sector i. Equation (4) indicates that even if one particular sector suffers high PER, the system PER may still meet the requirement as long as other sectors’PERs are low enough. On the other hand, the UAV may fail to collect enough data from one or some particular sectors of the whole circular area, although the lumped system performance seems to be acceptable.

Energy and operation models for UAV-WSN

We assume that the physical layer uses direct sequence CDMA with a spreading gain of K, and there are a total of N orthogonal codes available for the sensors. Each transmitting sensor selects one of these N codes to transmit its data. Since in most cases, there is a line-of-sight (LOS) communication between a sensor and the UAV, it is assumed that the signals only suffer from flat fading with the signal envelope satisfying the Rician distribution with a probability density function (pdf) given below⁸

ρ (r) = {\begin{matrix} \frac{r}{σ^{2}} e^{- \frac{(r^{2} + A^{2})}{2 σ^{2}} I_{0} (\frac{A_{r}}{σ^{2}})} \\ 0 \end{matrix} \begin{matrix} (A \geq 0, r \geq 0) \\ (r < 0) \end{matrix}

(5)

where A denotes the peak amplitude of the signal and $I_{0} (\cdot)$ is the zero-order Bessel function of the first kind. Let $γ_{i}$ represent the channel state between sensor i and the UAV.

According to LEACH,²⁴ the power consumed per second by a sensor node that transmits data d meters onward, denoted as p_tx(SN), is

p_{tx} (SN) = (α 1 + α 2 d^{th}) * Bits

(6)

and power consumed per second by a sensor node that receives data, denoted as p_rx(SN) is

p_{rx} (SN) = α 1 * Bits

(7)

Here, α1 is the power consumption for transmitter circuitry and α2 is the power consumption due to the amplifier. th is path loss exponent, and it takes the value of 2 or 4 in most cases. d is the distance between the node and the UAV. Bits is number of bits transmitted per second.

Albeit cycling through the two operating modes, the sleep mode and the work mode, to save energy, sensors are predominately in the sleep mode, and they will be woken up only after they receive a wake-up beacon signal from the UAV. After completing their data transmission to the UAV, nodes will quickly go back to the sleep mode again. Assume that the UAV’s antenna points downward when transmitting and receiving data packets and the sensors that can communicate with the UAV form a circle. Together the UAV and these sensors right below it form a communication cone. The sensors within a communication cone turn to be active and all of them are able to communicate with the UAV. Assume that the UAV contentiously send out beacon signals which serve dual purposes, synchronizing with the sensors and waking up sensors in the sleep mode. In essence, when a sensor receives a beacon, it transitions to the work mode and uses the received signal to estimate the uplink channel condition.

Transmission control

The uplink between the $m^{th}$ sensor node and the UAV during slot t is parameterized by its signal to noise ratio (SNR) $γ_{m}^{(t)}$ , that is independent and distributed with probability density f(γ), for m = 1, …, M and t = 0, 1, …. According to Venkitasubramaniam et al.,⁸ the maximum stable throughput λ is given by

λ (s (\cdot)) = {(1 - \int_{0}^{\infty} f (γ) s (γ) d γ)}^{M - 1} (\int_{γ_{0}}^{\infty} f (γ) s (γ) d γ)

(8)

where M is the number of the sensor nodes which have data to transmit, and $s (\cdot)$ is the scheduler. Denote $p_{γ_{0}} = P (γ \geq γ_{0})$ . If $p_{γ_{0}} \geq 1 / M$ , the optimal scheduler is

s^{*} (γ) = {\begin{matrix} 0 & γ < γ_{0} \\ \frac{1}{M p_{γ_{0}}} & γ \geq γ_{0} \end{matrix}

(9)

While for $p_{γ_{0}} < 1 / M$ , the optimal scheduler is

s^{*} (γ) = {\begin{matrix} 0 & γ < γ_{0} \\ 1 & γ \geq γ_{0} \end{matrix}

(10)

AO-Aloha: a CDMA-based MAC protocol for UAV-WSN systems

Packet formats

In AO-Aloha, there are four types of packets, namely, the Wake_Up packets, Sensor_Head packets, Slot_Allocate packets, and Sensor_Data packets, as shown in Figure 3. Serving as a beacon signal, a Wake_Up packet includes a 2-bit pkt_type field that is used to define the packet type. A Sensor_Head packet has a Sensor_ID field and three location fields (Loc_x, Loc_y, and Loc_z) that together geometrically determine the location of a sensor node. A Slot_Allocate packet can have multiple Sensor_ID fields and multiple TS_TIME fields, each of which is 32 bits long. Every TS_TIME field corresponds to the allocated time that a sensor node can send its data to the UAV. A Sensor-Data packet has one Sensor_ID field and one Sensor_Data field; the sensor node with its id specified in the Sensor_ID field puts its data to be transmitted into the Sensor_Data field.

Figure 3.

The formats of the four packet types in AO-Aloha.

Message flow and timing diagram

In UAV-WSN, increasing beacon interval leads to a larger guard time and longer idle listening due to clock drift. This is because the guard time has to be large enough to accommodate the synchronization errors, while these errors are affected by the frequency deviation (skew) between two clocks and thus accumulated over time. Therefore, a longer beacon interval will lead to larger synchronization errors and hence a larger guard time is necessary. It has to be avoided as UAV performing data collection is timing critical task.

To avoid this problem, beacon interval is not increased in our protocol compared with the one used in O-Aloha,⁸ as shown in Figure 4. Instead of two types of slots (the beacon slot and the data slot) as defined in O-Aloha in Figure 4(a), there are actually three types of slots in AO-Aloha, namely, the beacon slot, head slot, and data slot in Figure 4(b). For the proposed protocol, one data session is divided as multiple slots. Slot time is set to be equal to one time unit and slot t is assumed to occupy the time interval [t, t + 1). And one can see that head slots are inserted between a beacon slot and a data slot without increasing the beacon interval.

Figure 4.

A data transmission session undergoes several intervals: (a) O-Aloha and (b) AO-Aloha.

A data transmission session begins with a handshake. At the beginning of each slot, the UAV transmits a beacon, and this beacon is used by each sensor to estimate the down link (from the UAV to itself) propagation gain. In the same token, the uplink propagation gain from the sensor to the UAV can also be estimated. In the following head slot, the sensor node sends a Sensor_Head packet to request a slot that the sensor node can send its data. During the data transmission period, each sensor transmits its data with a scheduler of $s (γ_{i} (t))$ defined by equation (9), and $γ_{i} (t)$ characterizes the channel status from sensor node i to the UAV during slot t.

The detailed data transmission process is given as following, as shown in Figure 5. The UAV sends a few Wake_Up packets to all the sensor nodes within its communication cone. Those sensor nodes who received the Wake_Up packets designated to them will wake up and transition to the work mode from the sleep mode. Each of the sensor nodes in the work mode sends a Sensor_Head packet back to the UAV to request slots for data transmission. In response to a particular sensor node’s request, the UAV sends a slot allocation packet which includes a CDMA code that the sensor node can use for data transmission. After the sensor node receives this Slot_Allocate packet, it will send its Sensor_Data packets at the next available data slot. As sensor nodes have been assigned unique CDMA codes through Slot_Allocate packets, data collisions can be avoided when sensor nodes send their data in the data slots. After a sensor node finishes sending its Sensor_Data packets, it returns to the sleep mode.

Figure 5.

Data transmission process.

Priority mechanism and admission control

In the AO-Aloha protocol, a queue is used to store the IDs of the sensor nodes residing in the same sector, and every queue is assigned with a priority based on its own counter’s value. The counter of a queue counts the number of the sensor nodes in a sector that have sent head packets to the UAV; priority of a queue is set as the reciprocal of its counter value, indicating that higher (lower) priority is assumed for the sensors that belong to the sectors with a smaller (larger) number of sensor nodes. As time goes by, the priorities can be altered to better reflect the traffic condition in the network. When the UAV needs to assign data slots to the sensor nodes, it selects the sensor nodes from the different queues according to their priorities in descending order.

Figure 6 shows the process how sensor nodes are admitted into the queues. As the communication area is divided into k sectors, there will be a total of k queues, and these k queues are given different priorities according to the values of their corresponding k counters. After a beacon signal is sent out by the UAV, all the counters are reset to 0. When the UAV receives a head packet from a sensor node, it checks the packet’s ID field and the location fields. Then, the UAV puts the ID of the sensor node into the right queue according to its location, after which the counter of the queue is increased by 1. The UAV next sends a slot allocation packet to the sensor node. The price to pay here is that the UAV then needs to keep longer queues and has to use more channels for data transmission.

Figure 6.

Flow diagram regarding how the ID of a sensor node is inserted into a queue in the UAV.

To avoid this problem, call-blocking method²⁵ is used:

When a head packet from a sensor node as a call is arrived, UAV may not find space for its ID in the full queue. Therefore, the call has to compete with new calls for the free channels. However, network congestion can occur due to too many arriving calls, which can result in the call still being not served.

New-Call-Blocking: If a new call cannot find free channels, it will be immediately blocked.

Performance analysis

In this section, performance of the proposed protocol measured by throughput and PERs is discussed.

Throughput

System throughput is a function of channel condition. The unconditional probability of transmission is given by

p_{s} = \int s (γ) f (γ) d γ

(11)

where s(γ) is the scheduler of throughput $γ$ , and f(γ) is probability density of throughput $γ$ . Assume that there are N sensors that need to send their head packets to the UAV at the beginning of interval t. These sensors randomly choose one of the $K_{code}$ orthogonal codes with the probability of a particular code being selected is $q = 1 / K_{code}$ . The number of head packets received by the UAV in the first head slot of the interval t can be given by Venkitasubramaniam et al.⁸

λ_{1} = \sum_{k = 1}^{N} (\begin{matrix} N \\ k \end{matrix}) {(1 - q p_{s})}^{N - k} {(q p_{s})}^{k} E (N_{k})

(12)

where $E (N_{k})$ is the expectation of the number of packets received by the UAV from k sensor nodes. Then, the number of head packets received by the UAV in the next head slot is given by

λ_{2} = \sum_{k = 1}^{N - λ_{1}} (\begin{matrix} λ_{1} \\ k \end{matrix}) {(1 - q p_{s})}^{λ_{1} - k} {(q p_{s})}^{k} E (N_{k})

(13)

Assume there are m head slots in a beacon slot, and then the number of head packets received by the UAV in the mth head slot is given as

λ_{i} = \sum_{k = 1}^{N - λ_{i - 1}} (\begin{matrix} λ_{i - 1} \\ k \end{matrix}) {(1 - q p_{s})}^{λ_{i - 1} - k} {(q p_{s})}^{k} E (N_{k}) (1 \leq i \leq m)

(14)

Sensors can randomly choose one of the orthogonal codes $K_{code}$ , and the number of packets received at the UAV is given as

λ = \frac{K_{code} \times l_{D}}{m \times (l_{H} + l_{B}) + l_{D}}

(15)

where l_H, l_B, and l_D are the respective lengths of the head slot, the beacon slot, and the data slot. Note that the length of each interval is $l_{I} = m \times (l_{H} + l_{B}) + l_{D}$ , and when l_H + l_B << l_D, according to equation (15), the number of packets received at the UAV approaches its maximum, K_code.

Distribution of the sensors

At the ith time slot, of the n sectors, the average number of the sensors that have successfully transmit data to the UAV in each sector is $N_{s_{i}} / n$ . The number of the sensor nodes that successfully transmit data in sector j is $N_{s_{i}}^{(j)}$ . Then, the variance of the number of successful sensors in the ith time slot is determined as

σ_{i}^{2} = \sum_{j = 1}^{n} {(N_{s_{i}}^{j} - \frac{N_{s_{i}}}{n})}^{2}

(16)

The expectation of this variance over time T given below is a good measure of the uniformity of the sensor data collection. That is, the smaller the value of the expectation, the more the uniform of the distribution of the data collected

E (σ^{2}) = \int_{0}^{T} σ_{i}^{2} di

(17)

Simulation result

In our experiments, we randomly distribute 50–150 sensor nodes in an area of 350 mm×350 mm. All the simulation parameters are summarized in Table 1. Each data frame is divided into five slots for sensor nodes to send their data. The system performance will be evaluated in terms of bit error rate, collision rate, and packet loss rate.

Table 1.

Simulation parameters.

Parameters	Values
Area (m²)	350 × 350
Number of sensors	50–150
UAV’s path	Preplanned
UAV’s altitude (m)	45–135
UAV’s speed (m/s)	10
Beacon’s flare angle (°)	60
Transmission range (m)	50
Packet size (bytes)	25,100
Transmission interval (s)	0.1
Slots in each frame	5
Receiving threshold (dBm)	1E−10
Carrier frequency (Hz)	10
Bandwidth (MHz)	30
Modulation	BPSK
Time to transmit a data packet (s)	1
Time to transmit a beacon packet (s)	0.1
CDMA spread spectrum codes	4
Initial power of node	2 J
Power consumed by transmit circuitry of node²⁴	50 nJ/bit
Power consumed by amplifier of node²⁴	100 pJ/bit/m² (th = 2) or 0.001 pJ/bit/m⁴ (th = 4)

UAV: unmanned aerial vehicle; CDMA: code division multiple access; BPSK: binary phase shift keying.

A comparison is made between the proposed protocol and the best known UAV-WSN MAC protocol, the O-Aloha.⁸ Like AO-Aloha, each sensor in the O-Aloha network uses estimate of the wireless links between the UAV and itself to set up the probability of sending its data. O-Aloha, however, does not have the “handshaking mechanism” as AO-Aloha protocol does.

Figure 7 features two curves that represent the relationship between the delay and the throughput of the O-Aloha and AO-Aloha protocols, respectively, when there are a total of four CDMA codes available for data transmission. One can see that the throughputs of using the two protocols are nearly in par at 5 s. After around 55 s when the communications are stabilized, throughput of the AO-Aloha is 36.8% higher than that of the O-Aloha. It is also shown that AO-Aloha needs more time to reach a stable throughput than O-Aloha, as in AO-Aloha, extra time is needed for the handshaking between the UAV and the sensors. Because length of a SENSOR_HEAD packet is much shorter than that of a SENSOR_DATA packet, the portion of sending a head packet in one time interval is small. Although more time is spent at the beacon slot in AO-Aloha than O-Aloha, throughput can quickly pick up in the next data slot. The number of packets for per second received at the UAV is less than 3. These results agree well with the performance estimate as predicted by equation (14) given in section “Performance analysis.”

Figure 7.

The throughputs of UAV-WSN systems using the two MAC protocols; the number of CDMA codes is 4.

When there are more CDMA codes, AO-Aloha needs to run through a longer transient time to reach its throughput equilibrium. Even so, as shown in Figure 8, where the number of the CDMA codes is set to 8, throughput of using AO-Aloha is still 39.5% higher than that of using O-Aloha.

Figure 8.

Throughputs of the two protocols when the number of CDMA code is 8.

In Figure 9, the number of the sensor nodes is set to be 100, twice as that in Figures 7 and 8. Compared with Figures 7 and 8, it can be seen that when the number of the sensors increases, the throughput of AO-Aloha is more than doubled as that of O-Aloha.

Figure 9.

The throughputs of the two protocols when the number of sensor nodes is 100.

As the number of sensor nodes increases, the probability of data collisions in the communication channel is higher. The results have demonstrated that the AO-Aloha protocol can effectively avoid most of the data collisions when compared with O-Aloha.

In Figure 10, the BERs of the system are presented when the number of sensors is 50, 100, and 150, respectively. It can be seen that, when the number of sensors is small, the BER of the whole system is small, and the system BER increases with the increase in the number of sensors. When there are 100 sensors, the system BER level goes up by 20%. Same trend can be observed for the case when the number of sensors is 150. The BER is quite large when the UAV just enters the sensor area. After some time, when the UAV deeply penetrates into the sensor area, the BER becomes stabilized.

Figure 10.

The BER comparison in different number of the sensors.

In Figure 11, the BERs are given for the cases when the flight heights are 45, 90, and 135 m, respectively. The number of sensors in all these cases is set to be 150. The results have shown that the BER increases with the increase in the UAV height. The flight height has greater influence on the BER than the number of sensor nodes.

Figure 11.

The BER comparison in different heights of the sensors.

We also ran an experiment to test uniformity of data collection. In this experiment, we randomly distribute 420 sensor nodes over an area of 450 mm×450 mm, as shown in Figure 12. Apparently, certain regions are more densely populated than the others. Out of the 420 sensor nodes, according to the simulation, 320 sensor nodes have successively transmitted data to the UAV, as shown in Figure 13. Those that did not send data packets are absent from Figure 13 for clarity. One can see that these 320 nodes are fairly uniformly distributed, as evidenced by the three randomly drawn cycles with an average node count of 6.

Figure 12.

Distribution of the sensor nodes (420 in total).

Figure 13.

Sensor nodes that have transmitted data to the UAV (320 in total).

In Figure 14, average residual power of node versus number of rounds by AO-Aloha and O-Aloha are given. For instance, in the first round, average residual power of node by AO-Aloha and O-Aloha is 0.5 and 0.38 J, respectively. Overall speaking, power consumption of AO-Aloha is 46% less than that of O-Aloha as AO-Aloha can be adaptively selected to perform data collection according to the sensor density.

Figure 14.

Average residual power of nodes versus number of rounds.

Conclusion

In this article, AO-Aloha, a CDMA-based MAC protocol for UAV-WSN systems, was proposed, which considers the uniformity of data collection, energy consumption, and transmission efficiency. Priority-based data collection mechanism ensures that uniform data collection across the area is covered by the ground sensor nodes that are typically distributed in a random fashion. To improve energy efficiency, the sensor nodes are put into the sleep mode when they do not need to send data to the UAV, and they will be shortly woken up by a beacon signal from the UAV for data transmission. System throughput is enhanced by including a handshake into every time interval of transmission. Simulation results have confirmed that our protocol has a significant higher throughput (30% or higher) than O-Aloha protocol under different scenarios.

Footnotes

Academic Editor: Suat Ozdemir

Declaration of conflicting interests

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

Funding

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

References

Lucchi

Giorgetti

Win

. Using a UAV to collect data from low-power wireless sensors. In: Proceedings of the XIX congresso nazionale (AIDAA), Forli, 17–21 September 2007, vol. 86, pp.141–150.

Sotheara

Aso

Aomi

. Effective data gathering and energy efficient communication protocol in wireless sensor networks employing UAV. In: Proceedings of the IEEE wireless communications and networking conference (WCNC), Istanbul, Turkey, 6–9 April 2014, pp.2342–2347. New York: IEEE.

Heidemann

Estrin

. An energy-efficient MAC protocol for wireless sensor networks. In: Proceedings of the twenty-first annual joint conference of the IEEE computer and communications societies (INFOCOM 2002), New York, 23–27 June 2002, pp.1567–1576. New York: IEEE.

Van Dam

Langendoen

. An adaptive energy-efficient MAC protocol for wireless sensor networks. In: Proceedings of the 1st international conference on embedded networked sensor systems (SenSys 2003), Los Angeles, CA, 5–7 November 2003, pp.171–180. New York: ACM.

Saha

Johnson

. RMAC: a routing-enhanced duty-cycle MAC protocol for wireless sensor networks. In: Proceedings of the 26th IEEE international conference on computer communications (INFOCOM 2007), Anchorage, AK, 6–12 May 2007, pp.1478–1486. New York: IEEE.

Tong

Tang

Xie

. P-MAC: a cross-layer duty cycle MAC protocol towards pipelining for wireless sensor networks. In: Proceedings of the 2011 IEEE international conference on communications (ICC), Kyoto, Japan, 5–9 June 2011, vol. 2007, pp.1–5. New York: IEEE.

Park

Shimamoto

. Power consumption and BER tradeoff of MAC protocol for wireless sensor network employing UAV. In: Proceedings of the IEEE advanced technologies for communications (ATC), Ho Chi Minh City, Vietnam, 20–22 October 2010, pp.23–28. New York: IEEE.

Venkitasubramaniam

Adireddy

Tong

. Opportunistic ALOHA and cross layer design for sensor networks. In: Proceedings of the IEEE military communications conference (MILCOM), Boston, MA, 13–16 October 2003, pp.705–710. New York: IEEE.

Mergen

Zhao

Tong

Sensor networks with mobile access: energy and capacity considerations. IEEE T Commun 2006; 54(11): 2033–2044.

10.

Ranjan

Varma

Challenges and implementation on cross layer design for wireless sensor networks. Wireless Pers Commun 2016; 86(2): 1037–1060.

11.

Song

Hatzinakos

A cross-layer architecture of wireless sensor networks for target tracking. IEEE/ACM T Network 2007; 15(1): 145–158.

12.

Alikhani

Kunz

Hilaire

. A central-networked cross-layer design framework for wireless sensor networks. In: Proceedings of the 6th international wireless communications and mobile computing conference, Caen, 28 June–2 July 2010, pp.301–305. New York: ACM.

13.

Harthi

Borst

Whiting

Distributed adaptive algorithms for optimal opportunistic medium access. Mobile Netw Appl 2011; 16(2): 217–230.

14.

Soy

Özdemir

Bayrak

An autonomous packet transmission strategy for opportunistic wireless sensor networks. Int J Distrib Sens N 2012; 2012: 1–14.

15.

Sung

Bojanczyk

Cross-layer delay-constrained packet scheduling over multiuser wireless channels. In: Proceedings of the wireless telecommunications symposium, Prague, 22–24 April 2009, pp.1–6. New York: IEEE.

16.

Yessad

Bouallouche-Medjkoune

Aïssani

A cross-layer routing protocol for balancing energy consumption in wireless sensor networks. Wireless Pers Commun 2015; 81: 1303–1320.

17.

Huang

Xiao

Soltani

. The evolution of MAC protocols in wireless sensor networks: a survey. IEEE Commun Surv Tutor 2013; 15(1): 101–120.

18.

Khanafer

Guennoun

Mouftah

. A survey of beacon-enabled IEEE 802.15.4 MAC protocols in wireless sensor network. IEEE Commun Surv Tutor 2014; 16(2): 856–876.

19.

Wang

Minn

Dhahir

NA.

A distributed opportunistic access scheme and its application to OFDMA systems. IEEE T Commun 2009; 57(3): 738–746.

20.

Ibrahim

Ercetin

ElBatt

Stability analysis of slotted aloha with opportunistic RF energy harvesting. IEEE J Sel Area Comm 2016; 34(5): 1477–1490.

21.

Uddin

Rosenberg

Zhuang

. Joint routing and medium access control in fixed random access wireless multihop networks. IEEE/ACM T Network 2014; 22(1): 80–93.

22.

Chen

Liang

Motani

. A two-level MAC protocol strategy for opportunistic spectrum access in cognitive radio networks. IEEE T Veh Technol 2011; 60(5): 2164–2180.

23.

Park

Shimamoto

. QoS constraint with prioritized frame selection CDMA MAC protocol for WSN employing UAV. In: Proceedings of the IEEE wireless networking for unmanned aerial vehicles (GLOBECOM), Miami, FL, 6–10 December 2010, pp.1826–1830. New York: IEEE.

24.

Heinzelman

Chandrakasan

Balakrishnan

Energy-efficient routing protocols for wireless microsensor networks. In Proceedings of the 33rd Hawaii international conference on system sciences (HICSS), Maui, HI, 4–7 June 2000, pp.3005–3014. New York: IEEE.

25.

Toklu

Erdem

OA.

BSC-MAC: energy efficiency in wireless sensor networks with base station control. Comput Netw 2014; 59: 91–100.

Adaptive-opportunistic Aloha: A media access control protocol for unmanned aerial vehicle–wireless sensor network systems

Abstract

Keywords

Introduction

UAV-WSN system and models

Communication model for UAV-WSN systems

Packet error rate model for UAV-WSN systems

Energy and operation models for UAV-WSN

Transmission control

AO-Aloha: a CDMA-based MAC protocol for UAV-WSN systems

Packet formats

Message flow and timing diagram

Priority mechanism and admission control

Performance analysis

Throughput

Distribution of the sensors

Simulation result

Conclusion

Footnotes

Declaration of conflicting interests

Funding

References