A Synchronous Engine for Wireless Sensor Networks

Fernando Royo, Teresa Olivares and Luis Orozco-Barbosa

Albacete Research Institute of Informatics
University of Castilla La Mancha
02071 Albacete, SPAIN
[froyo, teresa, lorozco]@dsi.uclm.es

Abstract. The development and deployment of robust and highly populated

Wireless Sensor Networks require addressing a wide variety of challenges.
Among them, the synchronization and energy management of the nodes com-
posing the network have been identified as two challenges to overcome. In this
paper, we follow a cross-layer approach by defining a synchronous engine built
across the radio, MAC and routing layers. The design of our proposal has
been based on the results of our previous research efforts dealing with
various experimental platforms and experimental trials.

1 Introduction
Traditionally, an energy efficient wireless MAC protocol is a protocol that mini-
mizes idle listening and overhearing [1]. In addition, as any other MAC protocol, it
should keep to a minimum the number of collisions and the protocol overhead. Idle
listening is a dominant factor on energy waste in most sensor network applications.
The central approach to reducing energy lost to idle listening is to lower the radio
duty cycle by turning the radio off part of the time. Duty cycle is the ratio between
listen time and a full listen/sleep interval.
Radio duty cycling, where the radio is off by default but wakes up periodically to
participate in potential network communication, has received significant attention in
the literature. However, the duty-cycling benefits achieved in theory and simulations
have often not translated to practice. This can be attributed mainly to the problem of
time uncertainty between sensor nodes. If the sleep/wakeup schedules of nodes do not
intersect, the communication can not take place. Note that each sensor node may have
its own notion of time governed by its local clock. The approaches used by MAC
protocols to address this problem of time uncertainty determine their energy con-
sumption [8]. The lack of techniques to accurately estimate time uncertainty also
impacts the ability to deploy long-lived sensor network applications.
Recent studies on MAC protocols for sensor networks observe that there is not
clear trend indicating that medium access for sensor networks is converging towards
a unique best solution [13]. Besides, many of the protocols being introduced in the
literature have only been evaluated in simulation. In order to impact the market, a

Please use the following format when citing this chapter:

Royo, F., Olivares, T., Orozco-Barbosa, L., 2007, in IFIP International Federation for Information Processing, Volume
248, Wireless Sensor and Actor Networks, eds. L. Orozco-Barbosa, Olivares, T., Casado, R., Bermudez, A., (Boston:
Springer), pp. 107-118.
108 Fernando Royo, Teresa Olivares, and Luis Orozco-Barbosa

protocol not only must perform well in simulation; it must also integrate well within
the implementations of wireless sensor network protocol architectures
This paper presents SA-MAC (Synchronous after Awake MAC), a new MAC pro-
tocol for wireless sensor networks. Our objective is to address some of the major
issues currently present in this type of networks. We are particular interested on those
issues related to the lower layers, such as energy management, overhearing, packet
overhead or idle listening. Furthermore, the synchronization utilized in the protocol
denotes a cross layer view, fusing some characteristics from the network layer over
the MAC layer.
This article presents this new and robust protocol, as a component of a complete
network architecture and coexisting with other important protocols. The remainder of
this article is structured as follows. Section 2 reviews the state-of-the art on MAC
protocols for sensor networks. Section 3 shows our previous work with sensor net-
works, implementations and results. Section 4 describes our proposal. Section 5 con-
cludes the paper and outlines our future work.

2 Background
Medium access for sensor networks is a very active research area. The B-MAC
protocol has been an important input to the MAC protocols world for wireless sensor
network. We show in the following table the most important protocols for wireless
sensor networks in the last five years, and then we briefly explain the most important
differences between them. Existing works mainly focus on two directions: TDM A
and contention-based protocols. Two classes of contention-based protocols are those
that add schedules and those that employ channel sampling. All protocols control the
radio duty cycle to avoid energy waste in idle listening; some also take approaches to
avoid overhearing or add other optimizations [10]. The major advantage of schedul-
ing is that a sender knows a receiver's wakeup time and thus transmits efficiently.
However, the cost of listening for an entire contention interval is about ten times the
cost of channel sampling, thus the overhead in lightly used networks is higher then
Low Power Listening (LPL) based approaches [4].
S-MAC, Sensor Media Access Control [1, 2], is a contention based MAC protocol
that adds into the MAC layer power management, link-level retransmission, duplicate
packet suppression, hidden terminal avoidance using RTS/CTS, and link-quality
estimation. S-MAC periodically sleeps, wakes up, listens to the channel, and then
returns to sleep. S-MAC is designed to operate like a "black box"it is optimized for
a representative set of workloads. S-MAC does not have any hooks to change its duty
cycle or reconfigure its parameters; instead, it combines link, routing, organization,
synchronization, and fragmentation services into a single protocol. All of these ser-
vices may be used by network protocols. In the power management scheme in S-
MAC, each active period is of fixed size, 115 ms, with a variable sleep period. The
length of the sleep period dictates the duty cycle of S-MAC. At the beginning of each
active period, nodes within a cell exchange synchronization information. S-MAC
uses Adaptive Listen, which allows S-MAC to achieve lower latency when relaying
multihop traffic. By changing the duty cycle, S-MAC can trade energy for latency [3].
 A Synchronous Engine for Wireless Sensor Networks 109

Table 1. Mac Protocols

NAME YEAR UNIVERSITY CHARACTERISTICS
S-MAC 2002 USC, USA Scheduled Fixed duty cycle
T-MAC 2003 Delft, Holland Scheduled Adaptive duty cycle
WISE-MAC 2004 CSEM, Switzerland Synchronized preamble sampling
B-MAC 2004 UCB, USA Adaptive preamble sampling
(CCA & LPL)
Z-MAC 2005 NCSU, USA hybrid protocol (TDMA/CSMA)
2005 UM, USA Uncertainty-driven time synchro-
UB-MAC nization
B-MAC+ 2006 Pisa, Italy B-MAC wake-up preamble divi-
sion
SCP-MAC 2006 USC, USA LPL + Scheduling
Crankshaft 2007 Delft, Holland SCP-MAC for Dense WSN
SA-MAC 2007 UCLM, Spain Synchronization based packet and
take advantage it for routing

To handle load variations in time and location T-MAC, Timeout MAC [5], intro-
duces an adaptive duty cycle in a novel way: by dynamically ending the active part of
it. This reduces the amount of energy wasted on idle listening, in which nodes wait
for potentially incoming messages, while still maintaining a reasonable throughput.
T-MAC in variable workload uses one fifth the power of S-MAC. In homogeneous
workloads, T-MAC and S-MAC perform equally well. T-MAC suffers from the same
complexity and scaling problems of S-MAC [3].
WiseMAC (Wireless Sensor MAC) [5] is based on the preamble sampling tech-
nique. This technique consists in regularly sampling the medium to check for activity.
By sampling the medium, we mean listening to the radio channel for a short duration,
e.g. the duration of a modulation symbol. All sensor nodes in a network sample the
medium with the same constant period. Their relative sampling schedule offsets are
independent and constant. This technique provides a very low power consumption
when the channel is idle. The disadvantages of this technique are that the (long)
wake-up preambles cause a throughput limitation and large power consumption over-
head in reception. The novel idea introduced by WiseMAC consists in letting the
access point learn the sampling schedule of all sensor nodes. Knowing the sampling
schedule of the destination, the access point starts the transmission just at the right
time with a wake-up preamble of minimized duration. WiseMAC differs from previ-
ous research on ad-hoc sensor networks, mainly because it focus on an infrastructure
topology, and investigate how the unconstrained energy supply of the access point
can be exploited.
B-MAC, Berkeley Media Access Control [4], a carrier sense media access protocol
for wireless sensor networks that provides a flexible interface to obtain ultra low
power operation, effective collision avoidance, and high channel utilization, is moti-
vated by the needs of monitoring applications. B-MAC supports on-the-fly reconfigu-
110 Fernando Royo, Teresa Olivares, and Luis Orozco-Barbosa

ration and provides bidirectional interfaces for system services to optimize perform-
ance, whether it be for throughput, latency, or power conservation [3]. To achieve
low power goals and workload requirements, S-MAC is not only a link protocol, but
also a network and configuration protocol. Applications and services must rely on
policies internal to S-MAC to adjust operation as node and network conditions
change; such changes are opaque to the application. In contrast, the B-MAC protocol
contains a small core of media access functionality. B-MAC uses clear channel as-
sessment (CCA) and packet back offs for channel arbitration, link layer acknowledg-
ments for reliability and low power listening (LPL) for low power communication.
To achieve low power operation, B-MAC employs an adaptive preamble sampling
scheme to reduce duty cycle and minimize idle listening. WiseMAC meets many of
B-MAC goals except that it has no mechanism to reconfigure based on changing
demands from services using the protocol [4].
Z-MAC [6] is a hybrid protocol for wireless sensor network that combines the
strengths of TDMA and CSMA while offsetting their weaknesses. Like CSMA, Z-
MAC achieves high channel utilization and low-latency under low contention and
like TDMA, achieves high channel utilization under high contention and reduces
collision among two-hop neighbours at a low cost. Z-MAC has the setup phase in
which it runs the following operations in sequence: neighbour discovery, slot assign-
ment, local frame exchange and global time synchronization. These operations run
only once during the setup phase and does not run until a significant change in the
network topology occurs. Z-MAC uses the CCA and LPL features of B-MAC. Thus,
its energy efficiency is no better than B-MACs under low data applications.
UB-MAC, uncertainty B-MAC [7], integrate an uncertainty-driven time synchro-
nization scheme with B-MAC, and empirically demonstrate one o two orders of mag-
nitude reduction in the transmit energy consumption at a node with negligible impact
on the packet loss rate. B-MAC uses an asynchronous technique that involves no time
synchronization or clock estimation to deal with the time uncertainty. Other tech-
niques such as S-MAC and T-MAC use synchronized techniques where explicit time
synchronization beacons are transmitted periodically between neighbouring nodes.
This enables the transmitter to turn on the radio at the right moment, but the inability
to deal effectively with time varying changes in clock drift force these techniques to
re-synchronize frequently. In [7] the authors, first, experimentally obtained long time-
scale data sets both in indoor and outdoor settings for Berkeley mica2 motes. They
then performed a detailed characterization period on accuracy and energy require-
ments. Second, they used the results of the empirical analysis to design an adaptive
time synchronization protocol and, lastly, they developed a prototype implementation
on mica2 motes for sense-response applications (only with three nodes).
B-MAC+ [9] is an enhancement of B-MAC. The basic idea consists in replacing
the pattern of the wake-up preamble of B-MAC, with a new pattern that contains
information about the size of the remaining part of the preamble not yet transmitted.
This information can be used by receivers to avoid wait states during significant por-
tions of the preamble transmission time, i.e. going to sleep and waking up when the
data payload is actually transmitted. The wake-up preamble of B-MAC+ is obtained
by dividing into slots the wake-up preamble of B-MAC. The experiments were made
only with two Tmote nodes, one sender node and one receiver node connected to the
 A Synchronous Engine for Wireless Sensor Networks 111

USB port of a PC running a modified version of TOSBase. The results show statis-
tics about the time the radio must be on to receive a packet, and relates the collected
data with the analytical mean values of both B-MAC+ and B-MAC. B-MAC+ outper-
forms B-MAC, as the countdown packets technique reduces the waste of time in
receiving long wake-up preambles.
SCP-MAC, Scheduled Channel Polling [10], eliminates long preambles in LPL for
all transmissions, and is able to operate ultra-low duty cycles when traffic is light by
synchronizing the channel polling times. It has been designed with two main goals:
first, to push the duty cycle an order of magnitude lower than in practical with current
MAC protocols and second, to adapt to variable traffic loads common in many sensor
networks applications. To schedule coordinated transmission and listen periods is the
approach of S-MAC and T-MAC. The schedule determines when a node should listen
and when it should sleep. In S-MAC and T-MAC nodes adopt common schedules,
synchronizing with periodic control messages. A receiver only listen to brief conten-
tion nodes participating in data transfer remain awake after contention periods, while
others can then sleep. Overhead is due to schedule maintenance and listening during
contention intervals if there is nothing to send. Another technique is LPL presented in
WiseMAC and B-MAC. In LPL, nodes wake up very briefly to check channel activ-
ity without actually receiving data. The authors of SCP-MAC call this action channel
polling (polling refers only to each node sampling the channel to check for activity).
Unfortunately, current LPL-based protocols have some problems: the duty cycle is
limited to 1-2% because the polling frequency needs to balance the cost on sending
preambles and polling the channel; this balance between sender and receiver costs
makes LPL-based protocols very sensitive to tuning for an expected neighbourhood
size and traffic rate; finally, it is challenging to adapt LPL directly to newer radios
like 802.15.4, since the specification limits the preamble size.
SCP-MAC adopts channel polling from LPL approaches. However, unlike LPL,
SCP-MAC synchronizes the polling times of all neighbouring nodes. SCP-MAC
distributes schedules much as developed by S-MAC: each node broadcast its schedule
in a SYNC packet to its neighbours every synchronization period. The key is to dis-
cover the optimal synchronization period and wakeup tone length that minimizes the
overall energy consumption. Also SCP-MAC eliminates long preambles, so its energy
performance is not sensitive to varying traffics loads. The authors of SCP-MAC have
implemented the protocols in TinyOS over the mica2 motes with the CC1000 radio,
and to provide a clean comparison of LPL and scheduling, they implement SCP as a
layer over basic LPL. They also describe a preliminary port to micaZ motes with the
CC2420 radio supporting IEEE 802.15.4. The relative performance of SCP improves
on never, faster radios like the CC2420, while that of LPL degrades.
Crankshaft [11] is a MAC protocol specifically targeted at dense wireless sensor
networks. It employs node synchronization and offset wake-up schedules to combat
the main cause of inefficiency in dense networks: overhearing by neighbouring
nodes. Further energy savings are gained by using efficient channel polling and con-
tention resolution techniques. Crankshaft employs a mechanism of channel polling
very similar to the SCP-MAC.
112 Fernando Royo, Teresa Olivares, and Luis Orozco-Barbosa

3 Protocol Engineering

Protocol engineering becomes a key area of research enabling the development

and deployment of power efficient networks. Cross-layer protocol design and power
management have become two main approaches towards the development of such
networks.
In order to explore the capabilities of state-of the-art technology, we have con-
ducted a set of experimental trials [14]. The hardware equipment used throughout our
experiments consisted often micaZ nodes (MOTE-KIT2400)[6] and a Stargate node
used as base station. This latter node gathers the data collected from the ten sensor
nodes. The nodes were located throughout different locations in a building character-
ized by different environmental conditions: sun light conditions, number of people
normally working at a given lab, among others.
Our first experiment consisted in deploying them operating under the control of
the application provided by the manufacturer. This applications collects and sends the
data to the sink every eight seconds to the base station (high load scenario). We ob-
serve that the battery life span was 82 hours.
Due to the limited life span of the system, we did focus our research on the analy-
sis of the power consumption, in particular, on the transmission system and associated
protocols. It is well known that overhearing is the main source of power mismanage-
ment. We choose two key elements to improve the lifetime of the network:
To program a new application for data capture and data transmission to the base
station
To efficiently control the radio use. In a previous work, we have shown that the
radio device consumes as much as 65% of the overall energy [12]

Regarding the second, we focus on the MAC layer and the different MAC proto-
cols implemented for wireless sensor network: S-MAC, B-MAC or Wise-MAC [12].
All these protocols are different to the standard IEEE 802.15.4 [16]. They have im-
portant characteristics addressing the power consumption at the MAC layer of wire-
less sensor network. However, they have been developed for the mica2-based plat-
forms (CC1000 radio chip) [17]. We have the started by adapting the B-MAC to the
micaZ (CC2420 radio chip) [15].
B-MAC is able to reduce the idle-listening, i.e., the time that the node spends lis-
tening the channel. B-MAC requires that each node should wake up periodically to
verify the channel activity. In case of detecting activity, the node keeps sensing the
channel. On the contrary, no activity is detected, it falls asleep. The time between two
consecutive wake-up periods is fixed by the check interval. B-MAC defines eight
check intervals, each one corresponding to a different listening mode.
In order to ensure that all the packets are properly received, the packets are sent
with a preamble whose length is longer than the check interval. B-MAC defines 8
sizes of preamble, each one related to a different way of transmission, which the
protocol denominates transmit mode.
 A Synchronous Engine for Wireless Sensor Networks 113

Another advantage of B-MAC, with respect to S-MAC, is its modularity and

flexibility. B-MAC provides accessible interfaces to the upper layers, allowing the
upper layers to transmit or listen in various modes.
B-MAC has been originally designed over the CC1000 radio chip (mica2). There-
fore, the first task has been to analyze the main differences between the two radio
chip systems. These are listed in Table 2.

Table 2. Summary of main sensor radio chip characteristics.

Chipcon CO000 Chipcon CC2420
Chip UHF RF transceiver 2.4 Ghz IEEE 802.15.4 systems
Frecuency Range 300-1000 250 Kbps
Mhz
Integrated bit synchroniser Suitable for both RFD and FFD
Consumption: RX 9.6 mA TX Consumption: Rx 18.8 mA TX
26.7 mA 17.4 mA
Independent data buffering for
rx and tx
Encrypt (AES 128) |

As seen from the table, the use of the radio CC2420 provides some advantages to
micaZ nodes, characteristics as throughput or the possibility to encrypt, but the fact
that these new chip radio comply standards like IEEE 802.15.4, limits them in certain
aspects. The preamble length is fixed by the standard. This is the main obstacle to the
implementation of B-MAC on the CC2420 radio chip. B-MAC contains three impor-
tant interfaces MacControl, MacBackoff and Low PowerListening, although the two
early interfaces have already been implemented, the interface LowPowerListening
(LPL) has been until today an insurmountable step.
Our proposal, according to the recommendations of the protocol designer, it is
break with the idea of, if sender is not capable of prolonging the preamble for the
total reception of the packet on the part of the receiver, that be the receiver that before
the packet arrival event maintains the radio on for its total reception (see Figure 1).
This is the main idea before the preamble problem, but will also be necessary to im-
plement the rest of the interface that permit to level of application, and of transparent
way, to define the listening and transmit mode desired.

m
radio on
nratoon

Sender sleep | sleep

radio on ../*

Receiver sleep sleep

Foro* time
W*k*Up

Fig. l.On-off Timing.

114 Fernando Royo, Teresa Olivares, and Luis Orozco-Barbosa

The mechanism forcing the radio to keep awake should be incorporated into the
LPL interface, i.e., made available to the upper levels. Our objective is testing the
improvement using B-MAC and our own application, IntellBuildApp. We design
different tests to evaluate the lifetime of network and to collect data allowing us to
monitor the environmental conditions of the building. Four sets of trials were defined:

Test 1: Use of the application as defined by the manufacturers based on the

MTS400 and micaZ board. It sends a packet every 8 seconds, with all the data
from sensor board. It makes use of the IEEE 802.15.4 MAC protocol.
Test 2: Our own protocol. A packet is sent every 8 seconds, only temperature
and humidity data are sent. It uses a multihop routing protocol and the standard
802.15.4 MAC protocol.
Test 3: Same configuration as Test 2, but it uses the new interface implemented
over B-MAC, using listening mode 3 and transmit mode 4 (1200 ms radio on,
1100 ms radio sleep).
Test 4: Same configuration as Test 2, but it uses the new interface implemented
over B-MAC, using listening mode 5 and transmit mode 3(2500 ms radio on,
900 ms radio sleep).

We evaluated the performance of the different set-ups in terms of the network

lifetime and packet loss count. In this way, we can evaluate the energy-efficiency of
B-MAC for wireless sensor networks. In terms of the network lifetime, the best re-
sults were obtained when the listening mode was longer than the transmit mode. Fig-
ure 2 graphically shows the results obtained. However, Figure 3 shows that the packet
loss count is extremely high. This is due to the lack of synchronisation among the
nodes. While some nodes may attempt to transmit, the potential receivers may be
sleeping. It is clear that there is a need of a synchronisation scheme among the nodes.
A

Node Lifetime

I XMTS400

i IntellBuilApp

i IntellBuilApp
BMAC{3,4)

IntellBuilApp
8MAC(5,3)

& J& .J> _*&

Fig. 2. Lifetime of different nodes
 A Synchronous Engine for Wireless Sensor Networks 115

Lost Packets

XMTS400

I IntellDuilApp

m IntellBuilApp
BMAC(3,4)
IntellBuilApp
BMAC(5,3)

Fig. 3 Packet loss by node

4 SA-MAC: The Synchronous after Awake MAC engine

Bearing in mind the problem of packet loss and the need of developing an energy
efficient protocol, we have undertaken the design of SA-MAC. We consider the de-
sign of a node synchronization engine. This synchronization algorithm would manage
to save energy and establish routes, solving two issues in the area of wireless sensor
networks. To achieve the design goal, we developed SA-MAC that consists of two
phases: neighbour discovery phase and synchronization phase.
The main motivation towards the design of a synchronization scheme among
neighbouring nodes is to limit the idle listen. Since the wake-up procedure implies a
transient period, the synchronization should be designed by limiting the number of
wake up steps. Towards this end, all the children nodes to a parent should send one
after another before going back to sleep. In this way, the sink node only has to turn on
its radio for a shortest period of time.
In our proposal, the nodes exchange their schedules by broadcasting them to all
their immediate neighbours. It is also important to take into account the clock drift
between different nodes since this is a critical point in synchronization

4.1 Neighbour Discovery Phase

In this first phase, each node sends periodic neighbour discovery packets by
broadcasting them to all its immediate neighbours. In these packets, the node includes
the time, it is important for establishing the listening period for the parent node. In
our implementation, the discovery packets are sent every 20 seconds. If a node does
not receive any discovery packet from a node for two minutes, it is locally considered
inactive, i.e., no longer present.
When a node receives a neighbour discovery packet, it takes out the node id, its
send time and cost to send through it and save it in its neighbour table, see Figure 4.
Moreover, it adds an entry in its Planning Table.
116 Fernando Royo, Teresa Olivares, and Luis Orozco-Barbosa

Neighbour Table Planning Table

id send time active td cost stage delay

Fig. 4. Neighbour and Planning Tables.

For its election, it compares between the costs of the possible candidates, in this
first evaluation we work with hop number to the base station. Initially all entries set
up cost, stage, delay and active to not valid values, except for packets receive from
base station, this packet sets up its cost to 1, in such a way that this entry has priority
when is found for the scheduling algorithm and the node establishes it like parent
setting up stage to value parent. When a node has fixed its parent, it has to inform to
it for to be considered in the parent listen period. The node can also include the new
cost to arrive to the base station in its discovery packet. In this way the network to-
pology is generated as a tree, with the base station being the root with children nodes
in a first level (hop) and the remaining ones in successive levels.

4.2 Synchronization Phase

In this second phase, we describe how the node chosen as parent synchronizes
with all its children: As already stated, the children should send their packets one
after the other, remaining the father the minimum time in the listen state.

Parent node

Children nodel

Children node2

Children rxx)c3

Fig. 5. Synchronization phase.

This task begins when a node receives a packet from another node. The parent sets
up a record, in the Planning Table, for each child. In it, the parent indicates the time
delay for the associated child. This time delay is fixed in accordance with the number
of children associated to the father at that moment. We denote ttn as the time that the
parent node listens to each one of its children nodes and Nch(i) as the number of chil-
dren of a parent at time /. This time is transmitted by the father to each child node.

delaynode=ti* Nch(i)
 A Synchronous Engine for Wireless Sensor Networks 117

Once all children of a parent have received their delay, the parent sends a packet
via broadcasting to synchronize all its children nodes. Figure 8 depicts the ideal situa-
tion of a parent node synchronized with three nodes.
It is a dynamic process in which new nodes can be easily integrated. A new node
can be simply added to the end of the listening period. In this way SA-MAC is able
to adapt the topology of the network due to the addition of a new node. In a similar
way, nodes leaving the network can be simply discarded without requiring a long
signalling process.
The following code shows the activity of node when a packet is received.
receive(packet)
switch(packet_type)

case 'discovery':
if (node_is_known)
refresh_table();
else insert_to_table();
if(not_base_node)
planning_algorithm();
case 'information':
if(node_is_known)
insert_to_table() ;
refresh_table();
send_delay_to__children_node () ;
case 'synchronization':
save_delay();
case ' s t a r t ' :
send_event_to_application_layer_after_delay();

5 Conclusions and Future works

This paper introduces the implementation of synchronized MAC for CC2420 radio
chip. Energy efficiency is the primary goal in the protocol design. This is done by
addressing the key features of the MAC layer, such as, overhearing and idle listening.
By adopting a cross-layer design, we take advantage of the synchronization done at
the physical level for developing a routing strategy. We are currently working on the
development and testing of our proposal over an actual wireless sensor network.

References
[1] W. Ye, J. Heidemann and D. Estrin, An Energy-efficient MAC protocol for wire-
less sensor networks, In Proceedings of the 21st International Annual Joint Con-
ference of the IEEE Computer and Communications Societies (INFOCOM
2002), June 2002
118 Fernando Royo, Teresa Olivares, and Luis Orozco-Barbosa

[2] W. Ye, J. Heidemann and D. Estrin, Medium access control with coordinated,
adaptative sleeping for wireless sensor networks, In IEEE Transactions on Net-
working, April 2004.
[3] J. Polastre, A Unifying Link Abstraction for Wireless Sensor Networks, Doctoral
Dissertation, University of California, Berkeley .October 25, 2005.
[4] J. Polastre, J. Hill and D. Culler, Versatile Low Power Media Access for Wireless
Sensor Networks, In proceedings of the ACM SenSys'2004, November 2004,
Baltimore, Maryland, USA
[5] T. van Dam and K. Langendoen, An adaptive Energy-Efficient MAC Protocol for
Wireless Sensor Networks, In proceeding of the ACM Sensys'2003, November
2003, Los Angeles, California, USA
[6]I. Rhee, A. Warrier, M. Ahia and J. Min, Z-MAC: a Hybrid MAC for Wireless
Sensor Networks, In proceedings of the 3rd International Conference on Embed-
ded Networked Sensor Systems, 2005, San Diego, California
[7] S. Generiwal, D. Ganesan, H. Sim, V. Tsiatsis and M. Srivasta, Estimating Clock
Uncertainty for Efficient Duty-Cycling in Sensor Networks, In proceedings of
the ACM conference on Embedded Networked Sensor Systems, 2005, San
Diego, California
[8] V. Raghunathan, S. Generiwal and M. Srivastava, Emerging Techniques for Long
Lived Wireless Sensor Networks, IEEE Communications Magazine, April 2006
[9] M. Awenuti, P. Corsini, P. Masci and A. Vecchio, Increasing the efficiency of
preamble sampling protocols for wireless sensor networks, In proceedings of the
1st IEEE International conference on Mobile computing and Wireless Communi-
cations, Amman (Jordan), September 2006
[10] W. Ye, F. Silva and J. Heidemann, Ultra-Low Duty Cycle MAC with Scheduled
Channel Polling, In proceedings of the ACM conference on Embedded Net-
worked Sensor Systems, 2006, Boulder, Colorado, USA.
[11] G. P. Halkes and K.G. Langendoen, Crankshaft: An Energy-Efficient MAC-
Protocol for Dense Wireless Sensor Networks, In proceedings of the 4th Euro-
pean Conference EWSN 2007, Delft, The Netherlands
[12] T.Olivares, P.J.Tirado, L. Orozco-Barbosa, V. Lopez y P. Pedron: Simulation of
Power-aware Wireless Sensor Network Architectures, in ACM Internacional
Workshop on Performance Monitoring, Measurement, & Evaluation of Hetero-
geneous Wireless and Wired Networks. Torremolinos, Malaga (Espafia)
[13] K. Langendoen and G. Halkes, Embedded Systems Handbook, CRC Press, Aug.
2005
[14] T. Olivares, P.J. Tirado, F. Royo, J.C. Castillo and L. Orozco-Barbosa, Intell-
Building: A Wireless Sensor Network for Intelligent Buildings, In proceedings of
the 4th European Conference EWSN 2007, Delft, The Netherlands
[15] Crossbow Technology, INC MicaZ datasheet, www.xbow.com
[16] IEEE, Inc. Part 15.4: Wireless Medium Access Control (MAC) and Physical
Layer (PHY) Specifications for Low-Rate Wireless Personal Area Networks (LR-
WPANs), October 2003.
[17] Crossbow Technology, INC Mica2 datasheet, website, www.crossbow.com