Chapter 9: Time Synchronization

Chapter 9: Roadmap
 The time synchronization problem!
 Time synchronization in wireless sensor networks!
 Basic techniques for time synchronization!
 Time synchronization protocols!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
2!

Clocks and the Synchronization Problem

 Common time scale among sensor nodes is important for a variety of
reasons!
 identify causal relationships between events in the physical world!
 support the elimination of redundant data!
 facilitate sensor network operation and protocols!
 Typical clocks consist of quartz-stabilized oscillator and a counter that is
decremented with every oscillation of the quartz crystal!
 When counter reaches 0, it is reset to original value and interrupt is
generated!
 Each interrupt (clock tick) increments software clock (another counter)!
 Software clock can be read by applications using API!
 Software clock provides local time with C(t) being the clock reading at real
time t!
 Time resolution is the distance between two increments (ticks) of software
clock!
Fundamentals of Wireless Sensor Networks: Theory and Practice
Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
3!

1!
 Clock Parameters
 Clock offset: difference between the local times of two nodes!
 Synchronization is required to adjust clock readings such that they match!
 Clock rate: frequency at which a clock progresses!
 Clock skew: difference in frequencies of two clocks!
 Clock rate dC/dt depends on temperature, humidity, supply voltage, age of
quartz, etc., resulting in drift rate (dC/dt-1)!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
4!

Clock Parameters
 Maximum drift rate ρ given by manufacturer (typical 1ppm to 100ppm)!
 Guarantees that: dC
1− ρ ≤ ≤ 1+ ρ
dt

 Drift rate causes clocks to differ even after synchronization!

 Two synchronized
€ identical clocks can drift from each other at rate of at
most 2ρmax!
 To limit relative offset to δ seconds, the resynchronization interval τsync must
meet the requirement:!
δ
τ sync ≤
2 ρmax

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
5!

Clock Parameters
 C(t) must be piecewise continuous (strictly monotone function of time)!
 clock adjustments should occur gradually, e.g., using a linear
compensation function that changes the slope of the local time!
 simply jumping forward/backward in time can have unintended
consequences!
 time-triggered events may be repeated or skipped!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
6!

2!
 Time Synchronization
 External synchronization!
 clocks are synchronized with external source of time (reference clock)!
 reference clock is accurate real-time standard (e.g., UTC)!
 Internal synchronization!
 clocks are synchronized with each other (no support of reference clock)!
 goal is to obtain consistent view of time across all nodes in network!
 network-wide time may differ from external real-time standards!
 External synchronization also provides internal synchronization!
 Accuracy: maximum offset of a clock with respect to reference clock!
 Precision: maximum offset between any two clocks!
 If two nodes synchronized externally with accuracy of Δ, also synchronized
internally with precision 2Δ!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
7!

Why Time Synchronization in WSNs?

 Sensors in WSNs monitor objects and events in the physical world!
 Accurate temporal correlation is crucial to answer questions such as!
 how many moving objects have been detected?!
 what is the direction of the moving object?!
 what is the speed of the moving object?!
 If real-time ordering of events is t1<t2<t3, then sensor times should reflect
this ordering: C1(t1)<C2(t2)<C3(t3)!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
8!

Why Time Synchronization in WSNs?

 Time difference between sensor time stamps should correspond to real-time
differences: Δ=C2(t2)-C1(t1)=t2-t1!
 important for data fusion (aggregation of data from multiple sensors)!
 Synchronization needed by variety of applications and algorithms!
 communication protocols (at-most-once message delivery)!
 security (limit use of keys, detect replay attacks)!
 data consistency (caches, replicated data)!
 concurrency control (atomicity and mutual exclusion)!
 medium access control (accurate timing of channel access)!
 duty cycling (know when to sleep or wake up)!
 localization (based on techniques such as time-of-flight measurements)!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
9!

3!
 Challenges for Time Synchronization
 Traditional protocols (e.g., NTP) are designed for wired networks!
 WSNs pose a variety of additional challenges!
 Environmental effects!
 sensors often placed in harsh environments!
 fluctuations in temperature, pressure, humidity!
 Energy constraints!
 finite power sources (batteries)!
 time synchronization solutions should be energy-efficient!
 Wireless medium and mobility!
 throughput variations, error rates, radio interferences, asymmetric links!
 topology changes, density changes!
 node failure (battery depletion)!
 Other challenges!
 limited processor speeds or memory (lightweight algorithms)!
 cost and size of synchronization hardware (GPS)!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
10!

Synchronization Messages
 Pairwise synchronization: two nodes synchronize using at least one
message!
 Network-wide synchronization: repeat pairwise synchronization throughout
network!
 One-way message exchange:!
 single message containing a time stamp!
 difference can be obtained from (t2-t1)=D+δ (D=propagation delay)!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
11!

Synchronization Messages
 Two-way message exchange:!
 receiver node responds with message containing three time stamps!
 assumption: propagation delay is identical in both directions and clock
drift does not change between measurements!

(t 2 − t1 ) + (t 4 − t 3 ) (t 2 − t1 ) − (t 4 − t 3 )
D= offset =
2 2

€ €

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
12!

4!
 Receiver-Receiver Synchronization
 So far: sender-receiver approaches!
 Receiver-receiver: multiple receivers of broadcast messages exchange their
message arrival times to compute offsets among them!
 Example: 2 receivers; 3 messages (1 broadcast, 2 exchange messages)!
 No time stamp in broadcast message required!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
13!

Nondeterminism of Communication Latency

 Several components contribute to total communication latency!

 Send delay: !
 generation of synchronization message!
 passing message to network interface!
 includes delays caused by OS, network protocol stack, device driver!
 Access delay:!
 accessing the physical channel!
 mostly determined by medium access control (MAC) protocol!
 Propagation delay:!
 actual time for message to travel to sender (typically small)!
 Receive delay:!
 receiving and processing the message!
 notifying the host (e.g., interrupt)!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
14!

Nondeterminism of Communication Latency

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
15!

5!
 Reference Broadcasts
 Global Positioning System (GPS) is a well-known global source of time!
 time measured from epoch started at 0h January 6, 1980 UTC!
 unlike UTC, GPS not perturbed by leap seconds!
 GPS is ahead by 15 seconds (and increasing)!
 Terrestrial radio stations!
 WWV/WWVH & WWVB (National Institute of Standards & Technology)!
 continuously broadcast time based on atomic clocks!

 Problems with these techniques:!

 not universally available (underwater, indoors, outer space)!
 need for high-power receivers!
 size!
 cost!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
16!

Lightweight Tree-Based Synchronization

 Goal of LTS is to provide specified precision with little overhead!
 Based on pairwise synchronization:!
 message from j to k, containing time stamp t1 (jʼs clock)!
 message from k to j, containing t1 (jʼs clock) and t2, t3 (kʼs clock)!

 assuming message delay D!

t 2 − t 4 − t1 + t 3
offset =
2

€
Fundamentals of Wireless Sensor Networks: Theory and Practice
Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
17!

Lightweight Tree-Based Synchronization

 Centralized multi-hop version of LTS!
 reference node is root of spanning tree containing all nodes!
 breadth first search used to construct tree!
 once tree established, reference nodes synchronizes with children!
 errors from pairwise synchronization are additive!
 keep depth of tree small!
 overhead of pairwise synchronization: 3 messages!
 overhead of network-wide synchronization: 3n-3 messages (n edges) !

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
18!

6!
Lightweight Tree-Based Synchronization
 Distributed multi-hop version of LTS!
 one or more reference nodes contacted by sensors whenever
synchronization is required!
 nodes determine resynchronization period based on desired clock
accuracy, distance to reference node, clock drift ρ, time of last
synchronization!
 node can query neighbors for pending synchronization requests, i.e.,
node synchronizes with neighbor instead of reference node!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
19!

Timing-sync Protocol for Sensor Networks

 TPSN is another sender-receiver technique!

 Uses a tree to organize network!
 Uses two phases for synchronization!
 discovery phase!
 synchronization phase!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
20!

Timing-sync Protocol for Sensor Networks

 Level discovery phase!

 establish hierarchical topology!
 root resides at level 0!
 root initiates phase by broadcasting level_discovery message (contains
level and identity of sender)!
 receiver can determine own level (level of sender plus one)!
 receiver re-broadcasts message with its own identity and level!
 receiver discards multiple received broadcasts!
 if node does not know its level (corrupted messages, etc.), it can issue
level_request message to neighbors (which reply with their levels)!
 nodeʼs level is then one greater than the smallest level received!
 node failures can be handled similarly (i.e., if all neighbors at level
i-1 disappear, node issues level_request message!
 if root node dies, nodes in level 1 execute leader election algorithm!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
21!

7!
 Timing-sync Protocol for Sensor Networks

 Synchronization phase!
 pairwise synchronization along the edges of hierarchical structure!
 each node on level i synchronizes with nodes on level i-1!
 approach similar to LTS:!
– node j issues synchronization pulse at t1 (containing level and
time stamp)!
– node k receives message at t2 and responds with an ACK at t3
(containing t1, t2, t3, and level)!
– node j receives ACK at t4!
 node j calculates drift and propagation delay!
(t 2 − t1 ) + (t 4 − t 3 )
D=
2
(t − t ) − (t 4 − t 3 )
offset = 2 1
2

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
22!

Timing-sync Protocol for Sensor Networks

 Synchronization phase (contd.)!

 phase initiate by root node issuing time_sync packet!
 after waiting for random interval (to reduce contention), nodes in level 1
initiate two-way message exchange with root node!
 nodes on level 2 will overhear synchronization pulse and initiate two-
way message exchange with level 1 nodes after random delay!
 process continues throughout network!

 Synchronization error of TPSN!

 depth of hierarchical structure!
 end-to-end latencies!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
23!

Flooding Time Synchronization Protocol

 Goals of FTSP include:!
 network-wide synchronization with errors in microsecond range!
 scalability up to hundreds of nodes!
 robustness to topology changes!
 FTSP uses single broadcast message to establish synchronization points!
 Decomposes end-to-end delay into different components!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
24!

8!
 Flooding Time Synchronization Protocol
 t1: wireless radio informs CPU that it is ready for next message!
 d1: interrupt handling time (few microseconds)!
 t2: CPU generates time stamp!
 d2: encoding time (transform message into electromagnetic waves; deterministic, low hundreds of microseconds)!
 d3: propagation delay (from t3 on node i to t4 on node j; typically very small and deterministic)!
 d4: decoding time (deterministic, low hundreds of microseconds)!
 d5: byte alignment time (delay caused by different byte alignments (bit offsets), i.e., receiving radio has to
determine the offset from a known synchronization byte and then shift incoming message accordingly); can reach
several hundreds of microseconds!
 t7: interrupt, CPU obtains time stamp!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
25!

Flooding Time Synchronization Protocol

 Time-stamping in FTSP!
 sender sends single broadcast containing time stamp (estimated global time)!
 receiver extracts time stamp from message and time-stamps arrival (leads to
global-local time pair, providing a synchronization point)!
 synchronization message begins with preamble followed by SYNC bytes, data
field, and CRC!
 preamble bytes are used to synchronize receiver radio to carrier frequency!
 SYNC bytes are used to calculate bit offset!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
26!

Flooding Time Synchronization Protocol

 Time-stamping in FTSP (contd.)!
 multiple time stamps are used at both sender and receiver to reduce
jitter of interrupt handling and encoding/decoding times!
 time stamps are recorded at each byte boundary after the SYNC bytes
as they are transmitted or received!
 time stamps are normalized by subtracting appropriate integer multiple
of nominal byte transmission time (e.g., approx. 417μs on Mica2)!
 jitter in interrupt handling can be reduced by taking the minimum of
normalized time stamps!
 jitter in encoding/decoding can be reduced by averaging these
corrected normalized time stamps!
 final (error-corrected) time stamp is added into data part of message!
 at receiver side, time stamp must further be corrected by the byte
alignment time (can be determined from transmission speed and bit
offset)!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
27!

9!
 Flooding Time Synchronization Protocol
 Multi-hop synchronization!
 root node is elected based on unique node IDs!
 root node maintains global time and all other nodes synchronize to root!
 synchronization is triggered by broadcast message by the root node!
 whenever node does not receive synchronization message for
certain amount of time, it declares itself to be the new root!
 whenever root receives a message from node with lower node ID, it
gives up root status!
 all receiver nodes within range establish synchronization points!
 other nodes establish synchronization points from broadcasts of
synchronized nodes that are closer to the root!
 a new node joining the network with lowest node ID will first collect
synchronization messages to adjust its own clock before claiming root
status!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
28!

Reference-Broadcast Synchronization
 Key idea of RBS: in the wireless medium, broadcast messages will arrive at receivers
at approximately the same time!
 set of receivers synchronize with each other using a broadcast message!
 variability in message delay dominated by propagation delay and time needed to
receive and process incoming message (send delay and access delay are
identical)!
 RBS critical path is short than critical path of traditional techniques!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
29!

Reference-Broadcast Synchronization
 Example with 2 receivers:!
 receivers record arrival of synchronization message!
 receivers exchange recorded information!
 receivers calculate offset (difference of arrival times)!
 More than 2 receivers:!
 maximum phase error between all receiver pairs is expressed as group
dispersion!
 likelihood that a receiver is poorly synchronized increases with the
number of receivers (larger group dispersion)!
 increasing the number of broadcasts can reduce group dispersion!
 Offsets between two nodes can be computed as the average phase offsets
for all m packets received by receivers i and j:!
1 m
offset[i, j] = ∑ (T − Ti,k )
m k =1 j,k

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
30!

10!
 Reference-Broadcast Synchronization
 Multi-hop scenarios possible by establishing multiple reference beacons,
each with its own broadcast domain!
 Domains can overlap and nodes within overlapping regions serve as
bridges to allow synchronization across domains!

 RBS uses large amount of message exchanges!

 However, RBS is a good candidate for post-facto synchronization!
 nodes synchronize after event of interest has occurred to reconcile their
clocks!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
31!

Time-Diffusion Synchronization Protocol

 In TDP, nodes agree on network-wide equilibrium time and maintain clocks within a small
bounded deviation from this time!
 Nodes structure themselves into tree-like configuration with two roles:!
 master nodes!
 diffused leader nodes!
 TDPʼs Time Diffusion Procedure (TP) diffuses time information from master nodes to neighbors,
some of which become diffused leader nodes responsible for propagating the master nodeʼs
messages!
 During the active phase of TDP, master nodes are elected every τ seconds using an Election/
Reelection Procedure (ERP)!
 balances workload in the network!
 τ further divided into intervals of δ seconds, each beginning with the election of diffused
leader nodes!
 ERP eliminates leaf nodes and nodes with clocks that deviate from neighboring clocks by
more than a certain threshold (achieved through message exchanges to compare clocks)!
 ERP also considers energy status in election process!
 During the inactive phase of TDP, no time synchronization takes place!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
32!

Time-Diffusion Synchronization Protocol

 Elected master node broadcasts timing information to neighbors!
 Diffused leader nodes respond with ACK message!
 Master nodes determine round-trip delay Δj for each neighbor j, an estimate of one-way delay for
all neighbors (Δavg/2), and standard deviation of the round-trip delays!
 Standard deviation is sent in another time-stamped message to each neighboring diffused leader
node!
 Diffused leader nodes adjust their clocks using the time-stamp, the one-way delay estimation, and
the standard deviation!
 Diffused leader nodes repeat process with their neighbors (n times, where n is the distance from
the master node in hops)!
 Nodes receiving timing information messages from multiple masters use the standard deviations
as weighted ratio of their time contribution to the adjusted time!

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
33!

11!
 Mini-Sync and Tiny-Sync
 Pairwise synchronization approaches with low bandwidth, storage, and
processing requirements!
 Two clocks C1(t) and C2(t) can be represented as:!
C1 (t) = a12C2 (t) + b12
a12 = relative drift
b12 = relative offset
 Nodes can use two-way messaging scheme to determine these parameters:!
 node 1 sends a time-stamped probe at t0!
€
 node 2 responds with time-stamped replay at t1!
 node 1 records arrival time of reply at t2!
 results in data point (t0,t1,t2), which must satisfy:!
t 0 < a12 t1 + b12
t 2 > a12 t1 + b12

Fundamentals of Wireless Sensor Networks: Theory and Practice

Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
34!

Mini-Sync and Tiny-Sync

 Procedure is repeated several times to obtain series of data points and new
constraints on admissible values of a12 and b12!
 increases the precision of the algorithms!
 Not all data points are useful; every data point results in two constraints for
the relative drift and offset!
 Tiny-sync:!
 maintains only four of these constraints!
 whenever new data point available, the current 4 and the 2 new
constraints are compared and only the 4 that result in best estimates
are kept!
 disadvantage: constraints that may provide better estimates if combined
with other data points that have yet to occur may be eliminated!
 Mini-sync:!
 only discards a data point if it is certain that this point will be useless!
 larger computational and storage costs, but increased precision!
Fundamentals of Wireless Sensor Networks: Theory and Practice
Waltenegus Dargie and Christian Poellabauer © 2010 John Wiley & Sons Ltd.
35!

12!