See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/224601438

Wake-up receivers for wireless sensor networks: Benefits and challenges -

[Accepted from open call]

Article  in  IEEE Wireless Communications · September 2009

DOI: 10.1109/MWC.2009.5281260 · Source: IEEE Xplore

CITATIONS READS

146 1,472

3 authors:

Ilker Demirkol Cem Ersoy

Universitat Politècnica de Catalunya Bogazici University
81 PUBLICATIONS   2,511 CITATIONS    230 PUBLICATIONS   6,290 CITATIONS   

SEE PROFILE SEE PROFILE

Ertan Onur
Middle East Technical University
77 PUBLICATIONS   937 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Enhanced Mobility Management Mechanisms - 5G AURA project View project

Mobile Sink Routing View project

All content following this page was uploaded by Cem Ersoy on 11 March 2016.

The user has requested enhancement of the downloaded file.

 1

Wake-up Receivers for Wireless Sensor Networks:

Benefits and Challenges
Ilker Demirkol, Member, IEEE, Ertan Onur, Member, IEEE, Cem Ersoy, Senior Member, IEEE

Abstract—For successful data delivery, the destination nodes • Pure synchronous rendezvous: the sensor clocks are
should be listening the medium to receive the data when the pre-synchronized such that the wake-up time of each
sender node starts the data communication. To achieve this node is known a priori. This scheme requires recurrent
synchronization, there are different rendezvous schemes among
which the most energy-efficient scheme is utilizing wake-up time synchronization that consumes considerable energy.
receivers. Current hardware technologies of wake-up receivers Moreover, the sensors wake up even if there is no packet
enable us to evaluate them as a promising solution for wireless to transmit or receive which results in idle listening or
sensor networks (WSNs). In this work, the benefits achieved overhearing.
with wake-up receivers are investigated along with the challenges • Pseudo-asynchronous rendezvous: source nodes wake up
observed. In addition, an overview of state-of-the-art hardware
and networking protocol proposals presented. As the wake-up and emit a preamble signal that indicates the intention for
receivers offer new opportunities, new potential application areas data transmission. The preamble time is long enough to
are also presented and discussed. coincide with the wake-up schedule of the destination
Index Terms—Wake-up receiver, wireless sensor networks. node. Upon waking up and sensing the preamble the
destination node recognizes the intended packet transmis-
sion. In this scheme, time synchronization is not required,
I. I NTRODUCTION but sensors still follow a duty cycle and consume consid-
erable energy with preamble signalling.

W IRELESS sensor networks (WSNs) consist of au-

tonomous nodes that are equipped with sensors to
monitor the physical or environmental conditions. The objec-
• Pure asynchronous rendezvous: Sensors reside in deep
sleep and can be waken up by their neighbors on demand
with very low-power wake-up receivers. Whenever a
tive of a WSN is to sense the environment and to communicate node intends to send a packet, at first it wakes up
the collected information to the base station. Since energy the destination node, then sends the packet. Therefore,
is the bottleneck attribute of the WSNs, reducing redundant wake-up receivers are a solution to the redundant energy
energy consumption is a significant research issue. Instead of consumption caused by rendezvous.
sleep scheduling, on-demand or event-triggered operation of
The current wake-up receiver hardware products on the
the network will improve the performance and the lifetime.
market along with its ongoing hardware research enable us
This approach is viable with the wake-up receiver technology.
to evaluate the wake-up receiver technique as a candidate
In this paper, we address the benefits and the challenges of
feature for numerous WSN applications. In this paper, we
utilizing the wake-up receivers in the WSN applications.
investigate the wake-up receiver technology thoroughly with
In general, the energy consumed for communication is dom- its past, present and future. After presenting the state of the
inant compared to that for computation or sensing activities
art of the wake-up technology, we analyze the benefits and
of the sensors. Hence, the main objective in WSN research challenges and point out their application areas in the wireless
is to reduce the unnecessary communication. An efficient sensor networks.
protocol design must reduce the amount of communication
while achieving the requirements of the network. The factors
that impact the energy consumption of sensors in terms of II. S TATE OF THE A RT IN WAKE - UP R ECEIVER R ESEARCH
communication are shown in Table I. Studies on wake-up receivers consist of two parts: i) Hard-
One method to decrease the unnecessary energy consump- ware development for low-power wake-up circuitry and ii)
tion is to implement an energy-efficient configuration protocol Software development for networking protocols that utilize
where sensors operate under a well-defined sleep schedule. wake-up receivers. The wake-up receiver technologies pro-
That is, inactive sensors switch to off-mode or low-power posed in the literature or available on the market exhibit
listening mode. When two nodes are to communicate, the various working principles. We present the classification tree
receiver node must be awake when the sender initiates the of these wake-up receiver technologies that are applicable
communication which is referred as rendezvous [1]. There are to both hardware and software in Fig. 1. For the hardware
three types of rendezvous schemes: proposals and the products, this classification represents the
features offered, whereas for the protocol studies, this tree
I. Demirkol, E. Onur and C. Ersoy are with the Network Research represents the hardware features utilized.
Laboratory, Department of Computer Engineering, Boǧaziçi University, Bebek
34341 Istanbul, Turkey. e-mail:{ilker,onur,ersoy}@boun.edu.tr Based on their energy sources, wake-up receiver proposals
This work is supported by TUBITAK under the grant number 106E082. can be classified as passive wake-up receivers in which the
 2

TABLE I
T HE FACTORS THAT IMPACT THE ENERGY CONSUMPTION OF SENSORS CATEGORIZED BY THE OSI LAYERS

Application Layer · The data or packet redundancy such as correlated event detections
· The messaging caused by application requirements such as software updates
· The messaging done for monitoring activities such as health checking of the sensors
Transport Layer · The messaging done to provide end-to-end reliability, flow control and congestion avoidance
Network Layer · The idle listening caused by unnecessary wakeups of routing backbone nodes when there is no transmission intended
· Path augmentation, i.e., not using the shortest path to the sink (intentionally by the energy-aware routing methods or unintentionally
by the localized protocols)
· The messaging done for route establishment
· Point-to-point reliability solutions such as multi-path routing.
Data link Layer · Overhearing
· Idle listening
· Retransmissions because of collisions
Physical Layer · The employed physical layer technology such as narrow-band or spread-spectrum
· The employed methods to increase the robustness to interference
· The design decisions about the trade-off between bandwidth efficiency and energy consumption

TABLE II
Power Source Wake-up Signal Type
C URRENT CONSUMPTION VALUES FOR A WAKE - UP RECEIVER ON THE
MARKET [3]
Passive (External) Radio-based
Active (Internal) Acoustic Operating Mode Regulator on, Regulator off,
VCC = 3 V VCC = 2.4 V
Sleep 0.8 µA 0.3 µA
Wake-up Channel Destination Standby 7.0 µA 6.5 µA
Receive 7.2 µA 6.8 µA
Specification
Shared
Identity-based
Separate acoustic wake-up receivers are triggered by acoustic signals,
Range-based i.e. by external sounds. When the observed level of the external
Single channel
Multiple channels sound reaches a threshold, the wake-up circuitry is activated.

Fig. 1. Classification of wake-up receiver schemes A. Hardware Level Proposals

Commercialized wake-up receivers that can achieve low-
energy consumption by using an active wake-up receiver
wake-up circuitry is triggered by an external energy source exist. For instance, a wake-up receiver with three channels
and as active wake-up receivers in which the internal battery is available on the market whose simplified block diagram is
is used such that the wake-up receiver monitors for the possible shown in Fig. 2. It has three power management modes: sleep,
wake-up signal. The channel that the wake-up signal is sent standby and receive [3]. Table II presents the currents drawn at
can either be the same with the main radio communication each mode. In the sleep mode, all channels are switched off. In
channel, i.e., shared channel or a separate channel can be used the standby mode, selected channels are switched on and are
for the wake-up signalling. This separate wake-up channel may ready to receive data. The amplifier of the selected channel is
consists of multiple channels to be able to wake specific nodes on, whereas the correlator is powered down as long as no input
utilizing frequency division. Although a separate channel signal is detected at the input. The enabled channel switches
increases the cost and the complexity of the sensor node, as from the standby mode to the receive mode when the input
indicated in [2], the penalty of channel addition is minimal, signal is detected and stays in the receive mode as long as the
as the radio typically accounts for less than 15% of the cost input signal is detected. In the receive mode, the correlator of
of a sensor node. the channel is active and scans the input signal waveform for
The wake-up signal can be a single wake-up tone or a a valid wake-up pattern. The channel goes back to the standby
bit sequence. If all the nodes that receive the tone wakes mode after an adjustable timeout period if no input signal is
up, this scheme is referred as range-based wake-up which is detected. An optional regulator adjusts the internal voltage to
appropriate for multicasting. On the other hand, the wake-up 2.4 V. The current values presented in Table II are for the
signal may consist of a bit sequence to address the destination. case where all three channels are enabled. Further reduction
After reception of a wake-up signal, nodes check if the bit can be achieved with fewer channels. This wake-up receiver is
sequence refers to themselves. If so, the destination wakes up. triggered with a specific bit sequence, however, the wake-up
This scheme is referred as identity-based wake-up and allows pattern is not adjustable and hence only range-based wake-up
unicasting. can be achieved.
Radio signals are used as wake-up signals in radio-based Hardware proposals that are based on the super-regenerative
wake-up receivers, or shortly wake-up radio. Alternatively, principle are also available [4], [5]. The super-regenerative
 3

nodes so that when a transmission is done with a set of

Amplifier 1 frequencies, only the corresponding nodes wake up.
Correlator
RF Input 1
1
Other than the RF wake-up receivers, an acoustic wake-
up receiver is proposed in [8]. The acoustic wake-up re-
Amplifier 2
WAKE UP Signal ceiver consumes around 835 nW. Experiments are done with

lxer
Correlator

tile
Mup
RF Input 2
2 sounds generated by external devices and the level of the
Amplifier 3
sound determines the wake-up range. This technology can be
Correlator
improved to be embedded to the sensor nodes. Xbow Mica
RF Input 3
3
motes are equipped with a microphone that can be used as
an acoustic wake-up receiver, since a sounder also exists on
these nodes. Table III categorizes hardware proposals for the
Wake
Generator wake-up receivers based on the classification given in Fig. 1.

B. Software Level Proposals

Fig. 2. Simplified block diagram for a wake-up receiver on the market [3].
As the wake-up hardware improves and generates a promis-
ing solution for energy-efficient communication in wireless
receivers use a second lower frequency oscillator to provide sensor networks, related communication protocols are also
single device circuit gains of six order of magnitude. This enhanced. In this section, these studies are overviewed based
second oscillation periodically interrupts the main radio fre- on the classification presented in Fig. 1.
quency (RF) oscillation, allowing the RF signal to be build up 1) Communication protocols for active wake-up receivers:
continually. Hence, they enable very low-power active wake- A topology management technique that utilizes a wake-up
up receivers. An earlier work by Joehl et al. presents an im- radio is proposed in [2], namely Sparse Topology and Energy
plementation of a super-regenerative transceiver that consumes Management (STEM). The aim of the wake-up radio is to
3.6 mW for a receiver sensitivity of −105 dBm and the emitter separate data communication from the preamble messaging
current consumption is 6 mA for a 0 dBm output power [4]. used for the synchronization of the sender and receiver. Energy
Recently, a super-regenerative transceiver that consumes 400 saving is achieved by introducing latency at the synchro-
µW on reception and 1.6 mW on transmission is proposed for nization of the nodes via the wake-up radio. Even if the
wireless sensor networks by Otis et al. [5]. wake-up radio is not ultra-low power, energy savings can be
An important problem of low-power wake-up receivers is achieved with this method. Two variants exist where in STEM-
the false alarms caused by environmental noise. To overcome B, a beacon is sent which includes the identity of the sender
this difficulty, a wake-up receiver with three RF stages is and the destination node whereas in STEM-T, a busy-tone is
proposed in [6]. A simplistic stage consuming around nWs sent which wakes every node that hears the tone. STEM-B,
is always active. On receiving a signal, it activates a more therefore utilizes an identity-based wake-up receiver, whereas
complex stage consuming in µW to check if the received signal STEM-T utilizes range-based wake-up receiver.
is a specific wake-up code. Only if this stage confirms a valid Similarly, in [9], a wake-up receiver with duty cycle is
wake-up condition, the main transceiver is activated. Specific used which utilizes preamble based synchronization between
patterns of signals are used in a shared channel for wake-up the sender and the receiver. In addition to the wake-up with
signalling. preamble messaging, a triggered wake-up is also defined in
Gu et al. propose a novel approach in which the wake-up which every node wakes up once at duration T. Authors try
receiver is activated by the energy of the received radio signals to optimize the values of T based on the given packet arrival
[7]. This type of wake-up receivers can be categorized as rate with the aim of minimizing the total energy expenditure.
passive wake-up receivers since they do not require additional Since the wake-up delay increases the end-to-end delay of
power. To trigger the receiver, a period of time is required to data packets, a number of protocols are proposed to decrease
collect the adequate energy that is not negligible. The intro- the wake-up delay incurred. Latency minimized Energy Effi-
duced wake-up delay is proven analytically to be dependent cient MAC protocol (LEEM) is one such protocol which is a
on the wake-up range, i.e., the distance to the target node hop-ahead reservation scheme [10]. The idea is to reserve the
[7]. For instance, 55 ms delay is needed to wake up a node next hop’s channel, i.e., to wake-up the destination’s next hop
that is 30 meters away. The passive wake-up circuitry is not in advance.
totally battery-free, the comparators and amplifiers consume Another alternative to STEM is Pipelined Tone Wake-up
energy. Fortunately, the negligible current consumptions of Scheme (PTW) [11]. Again, wake-up receiver has a certain
the comparator and the amplifier are 350 nA and 880 nA, duty cycle. Range-based signaling is applied as in STEM-T.
respectively. The wake-up signals should be sent at a special Additionally, in PTW, the node that is receiving a data packet
radio frequency so that the normal radio communication does wakes up its neighbors for the next data transmission as in
not unintentionally wake up the idle part of the network. In LEEM.
addition, the authors present an identity-based wake-up with Wake-up receivers can be used for collision-free MAC-
the help of multiple transceivers with different frequencies. layer communication as utilized in [12]. Although, a routing
The combination of frequencies are assigned to the neighbor protocol is proposed, the underlying MAC protocol is defined
 4

TABLE III
C ATEGORIZATION OF WAKE - UP R ECEIVER H ARDWARE P ROPOSALS

Hardware Power Source Signal Type Wake-up Channel Destination Specification

AS3931 [3] Internal Radio-based Separate (Multi-Channel) Range-based
Otis et al. [5] Internal Radio-based Shared/Separate Range-based
Mark et al. [6] Internal Radio-based Shared Range-based
Gu et al. [7] External Radio-based Separate (Multi-Channel) Identity-based
Goldberg et al. [8] Internal Acoustic Separate Range-based

to send a wake-up signal on a broadcast channel. The address monitoring of its next hop before communicating with it. They
of the destination node is modulated with the wake-up. Access show that by utilizing the wake-up receivers, the same level of
to the broadcast channel is CSMA/CA. A similar approach security could be achieved whereas the total energy consumed
is applied in [13] where authors propose a MAC protocol is decreased by 66-90%.
which combines CSMA and CDMA techniques. The wake-up Table IV categorizes protocol studies that utilize wake-up
receiver used in [12] is identity-based whereas the one used receivers based on the classification given in Fig. 1.
in [13] is range-based.
Wake-up channel is used in [14] as a control channel where III. B ENEFITS , C HALLENGES AND T RADE - OFFS
RTS/CTS like messaging and busy-till information could be In this section, we present the benefits achieved with the
exchanged. Authors compare this protocol to S-MAC and utilization of the wake-up receivers and the challenges that
achieve nearly 66% reduction in energy consumption and 33- are observed in the wake-up receiver based WSN. These chal-
60% better end-to-end delay values. lenges should be studied carefully for the necessary solutions
Wake-up receiver idea is also proposed for other wireless to be able to take advantage of the offered benefits. Moreover,
networks. For instance, in [15], the lifetime of GSM devices the trade-offs observed in the networks with wake-up receivers
is improved using a low-power control channel on WLAN. are stated. Each of these trade-offs requires a detailed analysis
Experiment results show that the battery lifetime of these that enables the discovery of the optimum network settings.
devices can be improved by 115% with this methodology.
Lin et al. propose two asynchronous scheduling methods A. Benefits
and compare their performance results with the case of uti- • Energy-conservation: A dominant energy waste ob-
lizing wake-up radio [1]. They derive a lower bound for the served in WSN applications is the idle listening which
energy consumption of the wake-up radio which is achieved occurs because of periodical wake-ups regardless of the
by setting the probability that the sender node will estimate data communication needed. The wake-up strategy should
the destination node’s wake-up time within a specific time be energy efficient [20]. The wake-up receiver presents
threshold to 1. A similar performance comparison between the an energy-efficient solution to the idle listening as the
wake-up receiver and duty-cycled radio receiver is presented nodes only wake-up when there is intended message for
in [16]. them. Chowdhury et al. show that the energy consumption
In [17], analytical bounds on the lifetime of wireless sensor of S-MAC can be reduced significantly using wake-up
networks are studied. Authors utilize a wake-up radio in such receivers [14]. Since active wake-up receivers are also
a way that when a sensor wakes up and the channel is not an energy consumption source for the sensor nodes, they
free, it spends the awake time by switching to a low-power should be ultra low-power to acquire the energy gain.
wake-up radio. When the channel is free, the sensor powers • Minimum overhead: As the wake-up receiver presents
up its main radio to communicate. purely asynchronous rendezvous scheme, the overhead
Song et al. utilize a range-based wake-up receiver [18] for incurred by the time synchronization of the two other
a target tracking application on wireless sensor networks. The rendezvous schemes are dissipated. This overhead causes
node that detects the target first wakes up all its neighbors to higher energy consumption as shown by [1] and [16]. The
sense the target and to announce their observations. communication overhead to form the routing backbone
Various proposals presented in this section assume active also becomes redundant as each node wakes up its relay.
wake-up receivers which require internal source of energy. For instance, Dhanaraj et al. propose a method where a
However, the passive wake-up receivers that obtain their main destination node first wakes up its relay before receiving
energy from the external sources are also utilized for the the data intended for it [10]. In addition, instead of having
software proposals. clusters with regular duty cycles, formation of clusters
2) Communication protocols for passive wake-up receivers: can be done only with the event occurrences.
Khalil et al. propose Sleep-Wake Aware Local Monitoring • Different form of use: For efficient use of the WSN
(SLAM) [19] as a security mechanism for malicious nodes for a target application, wake-up receiver technology
in which a passive wake-up receiver utilized such as the enables various new methods such as the on-demand
one proposed in [7]. There are guard nodes that monitor the target monitoring method proposed in [18] where the
communications and try to figure out the malicious nodes. node that detects a target wakes up all its neighbors
Each node awakens the guard nodes responsible for local to receive their observations about the target. Section
 5

TABLE IV
C ATEGORIZATION OF WAKE - UP R ECEIVER BASED P ROTOCOL R ESEARCH

Protocol Power Source Signal Type Wake-up Channel Destination Specification

STEM [2] Internal Radio-based Separate (Single-Channel) Identity-based (STEM-B)/
Range-based (STEM-T)
Vaidya et al. [9] Internal Radio-based Separate (Single-Channel) Range-based
LEEM [10] Internal Radio-based Separate (Single-Channel) Range-based
PTW [11] Internal Radio-based Separate (Single-Channel) Range-based
Shah et al. [12] Internal Radio-based Separate (Single-Channel) Identity-based
Guo et al. [13] Internal Radio-based Separate (Single-Channel) Range-based
Chowdhury et al. [14] Internal Radio-based Separate (Single-Channel) Range-based
Shih et al. [15] Internal Radio-based Separate (Single-Channel) Identity-based
Lin et al. [1] Internal Radio-based Separate (Single-Channel) Range-based
Subramanian et al. [17] Internal Radio-based Shared Range-based
Song et al. [18] Internal Radio-based Shared Range-based
SLAM [19] External Radio-based Shared Range-based

IV presents application areas of the wake-up receiver form a connected network, the number of hops on the
technology. routing path may increase if the wake-up range is less
• Dynamic behavior: System-wide parameters create in- than the communication range which can cause inefficient
efficiencies as the network load differs based on the routing of the packets. Unfortunately, there is no work in
location. Hence, parameters such as the duty cycle, the the literature yet that investigates the effect of separate
preamble length should be dynamically set based on the topology caused by the wake-up receivers.
localized requirements. Since wake-up receiver enables • Energy Hole problem: Ahmed et al. define the hole
dynamic duty cycling and dynamic communication ren- problem as the result of some anomalies in the wireless
dezvous, several inefficient parameter definitions become sensor networks that impair the functionality of the
unnecessary that result in a more efficient network as network [21]. Specifically, the coverage hole is defined as
shown by [1] and [16]. the area not covered by any sensor, due to the anomalies
• Overhearing is decreased: With the help of identity- such as random deployment creating voids, node failures,
based wake-up receivers, overhearing is reduced since or jamming. If the wake-up range is smaller than the
only the destined nodes will wake up and listen the communication range of the main radio, the topology
medium as in [2] and [15]. of the network formed by the wake-up receivers may
contain more or larger energy holes. This may result in
B. Challenges inefficient network utilization, for instance to pass over
the larger energy holes requires more data communication
• Cost and Hardware Complexity: If the wake-up radio and hence more energy.
utilizes separate channel(s), then the hardware complexity • Overhearing increases for the shared channel and
and the cost of the sensor nodes will increase. This is range-based wake up receivers: For range-based wake-
especially important when multiple wake-up channels are up schemes, all the nodes hearing the tone switch to
employed as in [3] and [7]. However, as indicated in [2], the active state. Since all surrounding nodes will wake-
the radio typically accounts for less than 15% of the cost up at each transmission, overhearing can be increased
of a sensor node. compared to other protocols. To prevent the overhearing,
• Delay: With the passive wake-up radio, such as the identity-based wake-up receivers can be used as in [2]
one proposed in [7], wake-up delay is incurred which and [15].
may become significant for the requirements of the ap-
plication. The impact of the end-to-end delay may be
significant, when the number of hops to the sink is large. C. Trade-offs
For applications such as target tracking, point-to-point
(one-hop) delay is critical for the performance of the The following trade-offs are expected in the networks where
application. wake-up receivers are used. Although the quantification of
• Topology Change: Since the wake-up range may not be these trade-offs require detailed performance evaluation stud-
the same as the communication range of the main radio, ies, no work on this topic has appeared in the literature yet.
topology for the wake-up receivers can be different than • Wake-up range versus energy consumption: As the
the actual network topology. The neighborhoods achieved density of the sensor deployment increases, the distances
with wake-up receivers, hence, result in a different topol- among sensors become shorter. For dense deployments,
ogy than the original network topology. The former may multi-hop wake-up may be possible. In the multi-hop
result in a disconnected network, although the latter is wake-up scheme, a sensor node triggers a node that is
connected. In such circumstances, the problem should be several hops apart in the communication (routing) graph.
realized by the protocols and necessary solutions should For sparse deployments, the wake-up range affects the
be applied. In addition, even if the wake-up receivers energy consumption. The range for the wake-up radio can
 6

be adjusted with the energy consumption trade-off. Dis- For target tracking or sniper localization scenarios, the
tant sensors can be awaken by consuming more energy. sensors that detect an event may trigger the surrounding nodes
Determining the efficient wake-up distance depends on and a locally activated set of sensors may track/localize the
the design decisions such as the applied routing strategy. target. Activating only the sensors close to the path followed
• Wake-up range versus delay: Multi-hop communication by the target is a reasonable approach. The awaken group of
is common in WSNs. Hence, the end-to-end reporting sensors may localize the target more accurately. In surveillance
delay becomes a critical metric for the performance applications, the frontier sensors may invoke some set of
evaluation. Increasing the wake-up range, decreases the valuable or costly sensors behind the perimeter to produce
overall hop count which in turn decreases the delay. a barrier to reduce the false alarms. This scheme suggests a
• In-band or out-of-band wake-up radio: The same layered topology where overlays are probable.
channel(s) as that of the communication stack can be used The interaction of the wake-up receivers with the sen-
for the wake-up radio. This reduces the implementation sor node and with the network for a simple event driven
costs but has lower resistance to jamming attacks to the application is shown in Fig 3 based on a target detection
communication stack and increases the probability of col- scenario. The identity based wake-up scheme enables point-
lision. If separate channels are utilized the costs increase, to-point communication. While waking up the neighbors, it
however, resistance to jamming attacks increases and the is possible to communicate some valuable information. In the
energy consumption due to overhearing decreases. surveillance scenario shown in Fig 3, the sensor that detects
a target may communicate the type of the target (e.g., a
IV. A PPLICATION A REAS soldier or an armored military vehicle) which helps the awaken
When the WSNs are categorized according to the mode sensor decide on which sensing circuitry to activate. Such
of data collection, two main approaches are studied in the an application is reasonable if the sensing circuitry consumes
literature. These are event-driven and periodic (or query-based) comparable energy to the radio subsystem. For example, micro
data collection. In this section, we present how the wake-up impulse radar sensors spend a couple of orders higher energy
radio can be used for these application categories. compared to the passive infrared sensors. In this scenario,
when the first sensor detects the target, its sensing circuitry
triggers the wake-up receiver. First, a burst of signal is sent to
A. Event-driven Applications the next hop (second sensor) to trigger the first stage of the
In this category, sensors inform the sink about their de- wake-up receiver. Afterwards, the identity of the second sensor
cisions when a specific event occurs. Typical examples are and the target type are sent. The second stage of the wake-up
the forest fire detection system, sniper localization or the receiver of the second sensor turns on the main radio if the sent
surveillance system whose duty is intrusion detection, target identity matches itself. Then, the main radios can communicate
tracking or perimeter protection in scenarios such as border and the second sensor may turn on the neccessary sensing
monitoring against penetration by hostile elements. The events circuitry based on the type of the target.
in a WSN application can be spatiotemporally correlated. For A similar scenario can be devised to consider the preserva-
example, in a surveillance application, if a sensor detects a tion of the sensing coverage in case of sensor failures using
target, it is highly probable that another sensor at about the the redundancy of the deployment. Assume that each sensor
same distance will also detect the same target. Furthermore, is assigned one or more stand-by buddy sensors. When the
these detections will occur in about the same period of time. battery level of the active sensor drops below a certain thresh-
Due to the redundant deployment, the impact of the spatial and old, it can wake up one of its buddies to monitor the region.
temporal correlations can be very influential on the energy Two or more sensors that monitor almost the same area can
consumption. Instead of centralized decision fusion, a local be matched to act as the back-up. These scheme increases the
subset of sensors may collaborate to decide on the event using high-availability of the surveillance network through active-
data/information fusion techniques. If the sensors are equipped standby operation. A similar scenario is presented in [20].
with wake-up circuitry, a dynamic clustering mechanism can In an event-driven application, depending on the rate of
be viable instead of assigning sensors to clusters prior to the the events, utilization of the wake-up technology may become
network operation. Sensors that are spatially co-located with useless. If the events occur frequently, some set of sensors have
the event can form a cluster dynamically. This scheme may to sleep and wake up frequently. Hence, for such scenarios
not only yield better accuracy, but also consumes less energy. switching to periodic sleep scheduling is wiser. An example
For event-driven applications, the adaptivity of the network scenario can be the surveillance application of museums.
to some attributes of the event, such as its location, suggests Consider visitors of a historic arena. The objective of the WSN
developing cross-layer protocols with the sensing circuitry. is to detect the visitors whom are trying to enter unallowed
This is a challenging research topic that needs detailed analy- parts of the arena. During the open hours of the museum, the
sis. For example, based on the location of the event, the routing rate of the events may be larger compared to the night shifts.
strategy may be altered. In a surveillance application, assume Daytime operation of the network can be based on periodic
an intruder with a jammer. The sensor that detects the intruder, sleep scheduling, whereas the night time operation can follow
may apply a routing strategy to minimize the effect of the the wake-up scheme. Rigorous analysis of when to switch from
jammer by waking up the sensors that are apart enough from wake-up scheme to sleep scheduling is required when such bi-
the intruder. modal operation is probable.
 7

(3) Wake-up signal

is broadcasted with
Wake-up Antenna
the next hop ID and
type of the target
Main Antenna

First Stage Burst Detection (7) Message

(Detector, Comparator,Timer) is sent
Burst detection
Main
Radio
Subsystem
Second Stage Address Decoder
Code (ID) Sequence (LNA, Detector, Decoder)

Sensing Units
Wake-up Transmitter (1) Event occurred
(Oscillator, VGA, PA)
(2) Event sensed and Infrared MIR
wake-up signal
is to be sent
Sensor Node 1

LNA: Low Noise Amplifier

VGA: Variable Gain Adapter
PA: Power Amplifier The next-hop sensor is to be informed
MIR: Micro Impulse Radar

Wake-up Antenna

(4) Initial burst received, Main Antenna

timer set to receive code sequence
(7) Message
First Stage Burst Detection
is received
Burst detection (Detector, Comparator,Timer)
Main
Radio
(5) Code sequence received
Subsystem
Second Stage Address Decoder
Code (ID) Sequence (LNA, Detector, Decoder) (6) ID matches,
main radio
is activated

Sensing Units
Wake-up Transmitter
(Oscillator, VGA, PA)
Infrared MIR

Sensor Node 2

Fig. 3. Interaction of the wake-up receivers with the sensor node and with the network for a simple event driven application. The sensor node which detects
a target wakes up the next-hop sensor node. The wake-circuitry is simplified from [6].

B. Periodic or Query-based Applications sensors may reject to wake up. If a handshaking protocol is
utilized for the wake-up stage, the sensors may dynamically
Periodic or query based data collection is common when alter their route based on their impacts on the neighbors. For
environmental properties such as temperature or humidity example, during the handshaking process, the residual energy
levels of a field are to be monitored. For such applications, levels can be exchanged to minimize the effect of the energy
the wake-up circuitry may be utilized to dismiss the time consumption for the relay functionality.
synchronization requirement. Instead of scheduled operation,
a single sensor can manage the timing and when the data is to Assume that a WSN is deployed to monitor the depletion
be collected, it initiates the communication by waking up the of minerals in the soil where several types of crops are
neighbors. Hence, an individual sensor may create a dynamic cultivated. The farmer requires the average level of mineral
data flooding scheme where the data can be aggregated by ratios from a portion of the field where only a type of
the awaken sensors only. For example, the maximum value crop is nurtured. To obtain this information, the WSN must
of temperature readings can be calculated as the data flows to be queried based on the location of sensors. This type of
the sink through the dynamically determined route to the sink. sensor networks can be regarded as a distributed database.
Based on the residual energy levels of individual sensors, the For such applications, the dissemination of the queries requires
 8

intelligent routing techniques where the wake-up circuitry can [12] R. C. Shah and J. M. Rabaey, “Energy aware routing for low energy
be used. For example, in this farming scenario, the sensors ad hoc sensor networks,” in Proc. IEEE WCNC, vol. 1, Mar. 2002, pp.
350–355.
which are located in the region of interest can be awaken. The [13] C. Guo, L. C. Zhong, and J. Rabaey, “Low power distributed mac for
medium access layers can be coordinated during the wake-up ad hoc sensor radio networks,” in Proc. IEEE GLOBECOM, vol. 5, San
stage to avoid collisions during communication. Some packet Antonio, TX, USA, Nov. 2001, pp. 2944–2948.
[14] K. Chowdhury, N. Nandiraju, D. Cavalcanti, and D. Agrawal, “CMAC
transmission schedule for the specific event can be delivered to - A multi-channel energy efficient MAC for wireless sensor networks,”
the surrounding nodes within the wake-up code. For example, in Proc. IEEE WCNC, vol. 2, Apr. 2006, pp. 1172–1177.
SMAC can be modified to distribute the schedules when an [15] E. Shih, P. Bahl, and M. J. Sinclair, “Wake on wireless:: an event
driven energy saving strategy for battery operated devices,” in Proc.
event is detected. This minimizes the energy consumption for ACM MobiCom, Sep. 2002, pp. 160–171.
duty scheduling. [16] L. Zhong, J. M. Rabaey, and A. Wolisz, “An integrated data-link energy
model for wireless sensor networks,” in Proc. IEEE ICC, vol. 7, Jun.
V. C ONCLUSION 2004, pp. 3777– 3783.
[17] R. Subramanian and F. Fekri, “Sleep scheduling and lifetime maxi-
The wake-up receiver technology presents a promising mization in sensor networks: fundamental limits and optimal solutions,”
solution for the energy-efficient rendezvous between the sensor in IPSN ’06: Proceedings of the fifth international conference on
Information processing in sensor networks, Apr. 2006, pp. 218–225.
nodes. In this paper, a classification scheme is presented for the [18] L. Song and D. Hatzinakos, “A cross-layer architecture of wireless
wake-up receiver studies. The state-of-the-art wake-up receiver sensor networks for target tracking,” IEEE/ACM Trans. Netw., vol. 15,
hardware and networking protocol proposals are investigated no. 1, pp. 145–158, Feb. 2007.
[19] I. Khalil, S. Bagchi, and N. B. Shroff, “Slam: Sleep-wake aware local
based on this classification scheme. The benefits of the wake- monitoring in sensor networks,” in Proceedings of the 37th Annual
up receiver use is presented along with the challenges that has IEEE/IFIP International Conference on Dependable Systems and Net-
to be solved for its successful operation. In the systems with works, DSN ’07, 2007, pp. 565–574.
[20] A. Boukerche, X. Fei, and R. B. Araujo, “An optimal coverage-
wake-up receivers, a number of trade-offs are observed that has preserving scheme for wireless sensor networks based on local infor-
to be addressed for efficient network operation. These trade- mation exchange,” Computer Commununications, vol. 30, no. 14-15,
offs are also listed and discussed. The new WSN application pp. 2708–2720, 2007.
[21] N. Ahmed, S. S. Kanhere, and S. Jha, “The holes problem in wireless
possibilities that are made available with this new technology sensor networks: a survey,” Mobile Computing and Communications
are presented. Review, vol. 9, no. 2, pp. 4–18, 2005.

R EFERENCES
[1] E. Y. A. Lin, J. M. Rabaey, and A. Wolisz, “Power-efficient rendezvous
schemes for dense wireless sensor networks,” in Proc. IEEE ICC, vol. 7,
Jun. 2004, pp. 3769–3776.
[2] C. Schurgers, V. Tsiatsis, S. Ganeriwal, and M. Srivastava, “Optimizing
sensor networks in the energy-latency-density design space,” IEEE
Transactions on Mobile Computing, vol. 1, no. 1, pp. 70–80, Jan./Mar.
2002.
[3] 3D Low Power Wakeup Receiver-AS3931, Austria Microsystems, 2008.
[Online]. Available: http://www.austriamicrosystems.com/eng/content/
view/full/544
[4] N. Joehl, P. Dehollain, P. Favre, P. Deval, and M. Declerq, “A low-power
1-GHz super-regenerative transceiver with time-shared PLL control,”
IEEE Journal of Solid-State Circuits, vol. 36, no. 7, pp. 1025–1031,
Jul. 2001.
[5] B. Otis, Y. H. Chee, and J. Rabaey, “A 400µW-RX 1.6mW-TX
super-regenerative transceiver for wireless sensor networks,” in IEEE
International Solid-State Circuits Conference, ISSCC’05, Feb. 2005, pp.
396–7,606.
[6] S. von der Mark, R. Kamp, M. Huber, and G. Boeck, “Three stage
wakeup scheme for sensor networks,” in SBMO/IEEE MTT-S Interna-
tional Conference on Microwave and Optoelectronics, Jul. 2005, pp.
205–208.
[7] L. Gu and J. A. Stankovic, “Radio-triggered wake-up for wireless sensor
networks,” Real-Time Syst., vol. 29, no. 2-3, pp. 157–182, 2005.
[8] D. H. Goldberg, A. G. Andreou, P. Julián, P. O. Pouliquen, L. Riddle, and
R. Rosasco, “Vlsi implementation of an energy-aware wake-up detector
for an acoustic surveillance sensor network,” ACM Trans. Sen. Netw.,
vol. 2, no. 4, pp. 594–611, 2006.
[9] M. J. Miller and N. H. Vaidya, “A mac protocol to reduce sensor
network energy consumption using a wakeup radio,” IEEE Trans. Mobile
Comput., vol. 4, no. 3, pp. 228–242, May/Jun. 2005.
[10] M. Dhanaraj, B. S. Manoj, and C. S. R. Murthy, “A new energy efficient
protocol for minimizing multi-hop latency in wireless sensor networks,”
in PERCOM ’05: Proceedings of the Third IEEE International Confer-
ence on Pervasive Computing and Communications, Washington, DC,
USA, 2005, pp. 117–126.
[11] X. Yang and N. H.Vaidya, “A wakeup scheme for sensor networks:
Achieving balance between energy saving and end-to-end delay,” in
RTAS ’04: Proceedings of the 10th IEEE Real-Time and Embedded
Technology and Applications Symposium (RTAS’04), May 2004, pp. 19–
26.

View publication stats