Effective Audio Storage and Retrieval in Infrastructure less Environment over WSN

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056
Volume: 04 Issue: 03 | Mar -2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 1868
Effective Audio Storage and Retrieval in Infrastructure less
Environment over WSN
K.J.Anaghaa 1, V.Aishwarya2, Gayathri Suresh3, Vijaylakshmi .P4
1 K.J.Anaghaa, student, Panimalar Engineering College, Chennai, India
2 V.Aishwarya, student, Panimalar Engineering College, Chennai, India
3 Gayathri Suresh, student, Panimalar Engineering College, Chennai, India
4Vijaylakshmi .P, Associate Professor, Dept. of computer science and Engineering, Panimalar Engineering College,
Chennai, India.
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Audio represents one among the
foremost appealing yet however the least exploited
modalities in wireless sensingelementnetworks,because
of the possibly extremely giant information volumes that
have to be processedandits restrictedwireless capability.
Therefore, a way to effectively collect audio sensing data
remains a challenging drawback till date. During this
paper, we are proposing a brand new paradigm of audio
data assortment supported by the conception of audio-on
demand. We take into account a sink-free (no base
station) setting, targeting to help disaster management,
wherever audio chunks unit hold and store information
within the network for retrieval.
Key Words: Audio-storage,Audio retrieval,wireless
sensor network,WSN, Infrastructureless communication.
1 .INTRODUCTION
The rising wireless detector networks [1][[2] are
revolutionizing the ways that of assembling information
from the physical world [3]. The community has pictured an
outsized form of applications, such as setting watching,
scientific observation, underwater surveillance, and
structural health watching[4][5][6][7]. So far, audio
represents one in all the foremost appealing yet least
exploited modalities in detector networks, mainly because
high-frequency audio sampling will manufacture extremely
giant information volumes over bandwidth-limited links. In
this paper, we have a tendency to investigate a brand new
paradigm of audio services, specificallyaudio-on-demand,in
wireless detector networks. We take into account a
sink/infrastructure free setting targeting for earthquake
disaster management scenario, wherever any base
station/sink may well be broken during the disaster. We
have a tendency to decision our style in a sink-freeaudio-on-
demand WSN system. The target application is post-
earthquake search and rescue that becomes extremely
important with the hit of recent constant violent
earthquakes. When earthquake happens, recording and
storing acoustic events and providing an on-demand
retrieval service area unit is essential to the later rescue in
such systems. There are 2 main reasons for this: 1) The area
is commonly disconnected from the rest of the world, and 2)
Most of the acoustic events are units recorded before rescue
might happen.
Since individual sensors records are restricted in their
effective acoustic zones of geographic region, networks of
the acoustic sensors are required to enclose the affected
areas. The necessities of a reliableaudioon-demandservices
measure threefold.
1) Acoustic events ought to be recorded and kept within the
network, since the existent system might fail as a result
of calamity it could also be destroyed in ruinous
environmental conditions.
2) If there exists no base station or alternative
infrastructures, it’s tough hence not possible to efficiently
record or locate acoustic events.
3)The non-disruptive on-demand playback of the audio any
place within the network needs parallel knowledge
transferring and economical buffer pre-fetching
mechanism as a result of the restricted information
measure capability of WSNs.
With the recent advances in NAND nonvolatile storage, new
more prototypes area currently onthemarket thatinterface
Mica-class processing and radio hardware to up to 8GB of

flash memory [8]. The increasing in-network storage
capability indeed makes the on top of store-and-fetch
paradigm potential for WSNs, wherever the sensory
knowledge is hold on within the network and may be
retrieved on-demand. However, existing infrastructure-free
systems investigate solely the “store”[9] aspect ofthestore-
and-fetch paradigm, leaving the other aspect AN open
issue. For instance EnviroMic [10] employs a distributed
balanced storage mechanism to store the high-volume
sensory acoustic event knowledge within the network.
However, the sensory knowledge hold on within
EnviroMic can't be without delay accessedbeforeall sensors
are recovered from the experiment venue.
To support retrieval, a simple strategy is to locate the
info victimization flooding. The matter is that flooding
produces an oversized quantity oftraffic whereas thesearch
success rate can not be warranted. It’s natural to utilize
replication strategy to boost search potency. However, how
to achieve Associate in nursingoptimum replicationstrategy
with minimum retrieval energy consumption isn’t trivial for
audio applications, especially underneath Associate in
Nursing infrastructure-free andbandwidth-limited wireless
detector network. To deal with this problem, during
this paper we have a tendency to propose a
probabilistic thorough replication strategy. we have a
tendency to show that, by replicating each data replicas and
question replicas uniformly randomly across the network,
the projected strategy guarantees a high search successrate
with a determined boundary whereas the replication cost is
increased, wherever n is that the network size.
Based on the economical replication strategy, we tend
to implement the buffer pre-fetching moduleinThepropsed
system system, a real world sink-free audio-on demand
system over WSNs.In the propsed system uses time-division
cooperative recording technique for aggregation audio
sensory knowledge, wherever multiple sensors
detecting identical acoustic event type a bunch with AN
elected leader to assign the time slots. Thus, the nodes in
the cluster record the acoustic event hand and
glove successively.
Consequently, completely different chunks of and acoustic
event squaremeasure naturallyrecordedand holdon within
the style of time available audio chunks by completely
different nodes round the supply of the acoustic event in
several time slots. Insteadof replicatingrawaudiochunks,In
the propsed system encodes the information of chunks
residing on a node into a Bloom filter (BF) and replicates the
BF. A Bloom filter could be a space-efficient probabilistic
organization for representing a group. A BF can support
testing whether or not a component could be a member of a
group. The metadata of every chunk includes the
time available symbol and the location wherever the chunk
is hold on. Thus, we can compress thesetofchunksrecorded
by a node into a space efficient bit vector. By replicating the
bit vectors across the network, the propsed
system additional reduces the communication value for the
replication. Throughout retrieval, a question for a
selected chunk can be evaluated against the BFs. If matched,
the information chunks are often obtained from the origin
location. In SAoD [11] system with thirty IRIS motes
equipped with MTS310 detector boards. The experimental
results show that SAoD provides top quality audio-on
demand service with terribly slight
playback disturbance and shortstartuplatency.Results ofin
depth simulations in large scale networks show that SAoD’s
buffer pre-fetching will achieve bonded success
rate whereas reducing the energy cost by one order of
magnitude compared to existing theme. The main
contributions of this work square measure threefold:
 We have a tendency to style and implement a
true audio-on-demand system over WSNs
and appraise the performance using thirty nodes.
 We have a tendency to propose a completely
unique replication strategy that guarantees a high
chunk search success rate with a determined
lower bound at greatly reduced
communication value.
 We have a tendency to back any scale of the
communication value of replication by
cryptography the chunk data victimization Bloom
filter.
The rest of the paper is organized as follows. Section 2
reviews the connected work.. Section three describes the
detailed style of the proposed system. Section four presents
the system features and otheraspects. Sectionfive concludes
the paper with attainable extensions.
2. RELATED WORK
Most of the existing work on audio sensor networks focuses
on how to efficiently transfer the sensory data back toa base
station (sink) [12] by either using online stream
compression [13] or customizing high bandwidth sensor
prototype [14].In [12], Allen et al. deployed 16 sensor nodes
on the upper flanks of the Reventador active volcano to
collect the audio data. The nodes form a multi-hop routing

topology and relay data via a long-distance radio modem to
the observatory. They used the formed wireless sensor
network to continuously sample acoustic data at the active
volcano. A data collection protocol is designed to transfer
continuously sampled acoustic data to the base station. [15]
Tackled the problem of online compression of data streams
in a resource-constrained network environment, where
traditional compression techniques are not applicable.
Particularly, they aimed at fast piecewise linear
approximation methods with quality guarantee. They
studied two versions of the problem which explore quality
guarantees in different forms. For the error bounded
piecewise linear approximation problem, they designed a
fast online algorithms running in linear time complexity and
requiring a constantspacecost.Designedandimplementeda
high bandwidth system for quality-aware voice streaming
(QVS) in WSNs. QVS is built upon a new sensor hardware
platform for high-rate audio communication. In their design
they used the transceiver Chipcon CC1100 which has a 64
bytes FIFO buffer and maximum data rate of 500 kbps. They
used dynamicvoicecompressionandduplicationadaptation,
and distributed stream admission control techniques. Their
experimental results show that QVS delivers satisfactory
voice streaming quality.
The above existing work on audio services over WSNs
assumes the existence of a base station [16]. The
infrastructure-basedschemes,however,maybeproblematic
when applied totheaudio-on-demandapplicationaddressed
in this paper, because a user may hope to access onlylimited
audio events of interest from any place in the WSN just as
audio events are recorded everywhere. Transferring all the
sensory audio data to a single base station is costly and
infeasible. Moreover, a base station is a centralized point of
failure. The failure of a base station in a disaster will
paralyze the whole system. To the best ofourknowledge, we
are the first to design and implement an audio-on-demand
system over WSNs. The proposed retrieval schemebased on
replication is different from existing flooding and a
geographic hash table (GHT) [17]. Flooding does not
guarantee the success rate without exhaustively searching
all the sensor nodes.
The GHT partitions the name space over the nodes and has
good success rate for key-value search, while it suffers from
the problem of exact match. Furthermore, although the
problem of node failure for a key in GHT can be alleviated by
using more than one node for a key, GHT cannot survive the
catastrophic failure. However, the case that a large number
of nodes may be destroyed is normratherthantheexception
in the target application in this work.
3. SYSTEM DESIGN
In this section, we tend to gift the look of the propsed
system. We first briefly describe however the audio
events area unit recorded and stored. Then, we tend to
introduce however the propsed system replicates
the informationofaudiochunks withinthe compressed kind.
Finally, we describe the replicating and chunk
discovery theme.
Fig -1: Architecture Diagram
3.1 COOPERATIVE RECORDING
SAoD utilizes the cooperative recording theme planned in
[16] to separate the task of recording an acoustic event into
units divided by time slots among multiple sensors around
the acoustic event. Once multiple nodes discover identical
acoustic event at the same time, they type a gaggle. The
members of the cluster coordinate to elect a
pacesetter; United Nationsagency assignsrecordingtasksto
individual members successively. Thus, AN acoustic event
file, ai, is of course segmental on time. Audiois partitioned
off into chunks of uniform unit to make the file available on
time. Every chunk contains a fastenedplayingtimeadequate
the length of a time interval. Due to the limited
memory capability of sensing element nodes] and also
the serious loads caused by the data converter (ADC)
sampling, choosing a correct size of the slot isn't trivial. In
Section 5, we are goingto show however wehavea tendency
to get the best setting of time slots in the propsed system in
bigger detail. By victimization the cooperative recording
technique, the chunks of AN acoustic event filewill naturally
be collected by totally different sensor nodes
and keep during a distributed manner. Without the

cooperative recording technique, once an acoustic
event happens, an easy style can let all the nodes
which discover the event perform the ADC sampling to
record the event. Thus, identical event are going to
be recorded and keep by multiple sensing element nodes,
creating the theme expensive in each energy and storage
consumptions. The time-division cooperative
recording style will create audio chunks time
addressable. Thus in on every occasion slot, one chuck are
going to be recorded by one allotted node. Such
a style greatly reduces the redundancy of sampling and
storage. It also effectively achieves a far better load
balance. Throughout retrieval, different chunks are often
fetched from different nodes.
3.2 METADATA ENCODING
Instead of replicatingthe rawaudiochunks, wehave
a tendency to use Bloom filters [18] to write the
information of the chunks residing on a node. By replicating
the information in an exceedingly space-efficient means,
SAoD greatly reduces the communication value. The
Metadata is information about the data (Data about Data).
Metadata Encoding and Replication is replicating the
audio chunks and Bloom filter. We use Bloom filter to
encode the metadata of chunks residing on a node. By
replicating the metadata in a space efficient way in
greatly reduce the communication cost. The Bloom filter
is used to reduce the complexity of the searching. The
Bloom filter having a Hash map to store a sensor
recording time and sensors name and replicated node
information. To estimate a network size by use of a
vector. Each sensor transmitted to the own Bloom filter to
the other sensor in that networks.
3.3 NETWORK SIZE ESTIMATION
Without a base station, it's troublesome to get such
statistics [24]. To solve this downside, the propsed system
utilizes a variant of the gossip algorithm 1st projectedin[19]
to estimate the network size. The strong rule
allows each node to quickly collect the worldwide statistics
withinthe network.Initially, every node will befollowing the
experiment: it flips a coin up to l times and counts the
number of times the pinnacle seems before the primary tail.
It saves this count r in a very bit vector (all bits ab initio set
to 0) by setting the rth (counting from the correct end) little
bit of the vector to 1. Then the nodes within the system
network perform a gossip algorithm. Throughout every
spherical of gossip, every node arbitrarily selectsa neighbor
and sends its bit vector to the selected neighbor. The node
receiving the bit vector performs a bitwise-or operation
between the received bit vector and its native bitvector,and
replaces the native bit vector with the ensuing bit vector.
The strong gossip theme leads the computation
of aggregate data to converge exponentially: after rounds of
gossip, all nodes can get the calculable network size with
high chance. Moreover, the applied math worth of n is
roughly 2t1 0:77351 withhighprobability, wherever t isthat
the position of the primary “1” bit within the bit
vector tally from the left finish. In network formation the
sensors are to be created. Each sensor will have its own
memory and Battery. We are developing an effective audio
storing and retrieval in Infrastructure less environment.
So the Network is fixed sensor network. Each sensor is
fixed in a Disaster area and cover the maximum range
and sensing the information. The sensor neighbors are
calculated depending upon the coverage. If the maximum
number of sensor are created and fixed after that to
select a Header node in that network. If we choose a head
node based on sensor battery and memory because the
head should have a long life compared to other sensors.
3.4 REPLICA DEPLOYMENT
The best replication model requires that the replicas square
measure deployed every which way. In the propsed system
deployment, we have a tendency to assume that the sensors
square measure deployed uniformly at random during a
such as space. When the propsed system computes the best
variety of replicas to deploy consistent with Theorem 2, it
samples the best variety of random locations in the fastened
space and deploys the replicas of the data bit vector to the
nodes nearest to the situations.Beforecastingall thereplicas
to the set of chosen nodes, it forms a token spanning tree
among the haphazardly sampled/computed nodes.Here,the
logical neighbors within the token spanning tree
communicate with oneanothermistreatmenttheunderlying
geographic routing algorithms [20]. The replicas square
measure deployed using multicast on the token spanning
tree. The bright inexperienced nodes square measure those
with replicas deployed through the token spanning tree.
Note, the location information of the supply nodeisattached
with the Bloom filter for chunk downloading throughout
retrieval. In an algorithmic program one describes the data
replication strategy thoroughly.
3.5 Query Evaluation
During retrieval time query to the any sensor to
particular time sensed audio. The sensor check that time
audio is sensed to the current node or not. If the audio is
here then return to the sensed audio to the user
otherwise check the replicated Bloom filter. The

replicated Bloom filter contains sensor recorded time,
and that time recorded audio will be distributed the
particular sensors those sensors will be maintained at
Bloom filter. The replicated Bloom filter will be return to
the which sensors have the sensed audio and that sensor
should be return the sensed audio to the user. In case of
the recorded sensor should be damaged or destroyed due
to some problem. The Bloom filter will be returning the
alternative sensors and the sensors should be returning
the sensed audio to the user. Such that design achieves a
search success rate of 98 percent while reducing the
search energy consumption by an order of magnitude.
4. SYSTEM FEATURES
In a cluster, each monitored component is monitored by n
sensing nodes and it can communicate with each other
nodes. We assign the cluster name to each cluster and each
sensing node stores its cluster name. Each cluster can
communicate with the help of forwarding sensors. Each
sensing nodes can sense the data and forwardthedata to the
forwarding sensors. Then the measured data can be
forwarded to the controller with the help of forwarding
nodes. Each sensing node stores the stores the check
polynomial of other clusters. Data can be validated by using
this check polynomial.
En –route Filtering is an energy efficient scheme as the
false messages are filtered at intermediate nodes before
posing the impact on remaining nodes in the network. The
false message (or report) forged by compromised sensor
nodes can consume lots of network and computation
resources and shorten the lifetime of sensor networks.
Therefore, false reports should be filtered at forwarding
nodes as quickly as possible by using the secret key. In the
Real applications, to save energy the sensor nodes of the
propsed system are usually kept in the sleep mode [21].
5. CONCLUSIONS
In this project, work the project is to propose an
Effective Audio Storing and Retrieval in Infrastructure Less
Environment over WSN.There is an enhancement provided
for the Replicating of Audio chunks.Thatiswhentheaudiois
retrieved from the user it is provided in a single audio
format. If the sensor is damaged or destroyed the audio
chunks are retrieved from an alternative sensor.
6. REFERENCES
[1] H. M.Ammari and S. K. Das, “Centralized and clustered
k-coverage protocols for wireless sensor networks,” IEEE
Trans. Comput., vol. 61, no. 1, pp. 118–133, Jan. 2012.
[2] W. Huangfu, Z. Zhang, X. Chai, and K. Long, “Survive
ability oriented optimal node density for randomly
deployed wireless sensor networks,” Sci. China Inf. Sci.,
vol. 57, no. 2, pp. 1–6, 2014.
[3] X. Wen, L. Shao, Y. Xue, and W. Fang, “A rapid learning
algorithm for vehicle classification,” Inf. Sci., vol. 295, pp.
395–406, 2015.
[4] T. Zhang, D. Wang, J. Cao, Y. Q. Ni, L. Chen, and D.
Chen,“Elevator-assisted sensor data collection for
structural health monitoring,” IEEE Trans. Mobile
Comput., vol. 11, no. 10, pp. 1555–1568, Oct. 2012.
[5] Y. Li, C. Ai, C. T. Vu, Y. Pan, and R. A. Beyah, “Delay-
bounded and energy-efficient composite event monitoring
in heterogeneous wireless sensor networks,” IEEE Trans.
Parallel Distrib. Syst., vol. 21, no. 9, pp. 1373–1385, Sep.
2010.
[6] J. Xu, K. Li, and G. Min, “Reliable and energy-efficient
multipath communications in underwater sensor
networks,” IEEE Trans.Parallel Distrib. Syst., vol. 23, no. 7,
pp. 1326–1335, Jul. 2012.
[7] J. Shen, H. Tan, J. Wang, J. Wang, and S. Lee, “A novel
routing protocol providing good transmission relia bility
in underwater sensor networks,” J. Internet Technol., vol.
16, no. 1, pp. 171–178, 2015.
[8] K. Mihic, A. Mani, M. Rajashekhar, and P. Levis,
“MStore: Enabling storage-centric sensornet research,” in
Proc. ACM/IEEEInt. Conf. Inf. Process. Sens. Netw.,
Cambridge, MA, USA, Apr. 2007.
[9] J. Xu, X. Tang, and W.-C. Lee, “A new storage scheme for
approximate location queries in object-tracking sensor
networks,” IEEE Trans. Parallel Distrib. Syst., vol. 19, no. 2,
pp. 262–275, Feb. 2008.
[10] L. Luo, Q. Cao, C. Huang, T. Abdelzaher, J. A. Stankovic,
andM. Ward, “EnviroMic: Towards cooperative storage
and retrieval in audio sensor networks,” in Proc. 27th Int.
Conf. Distrib. Comput.Syst., 2007, p. 34.
[11] Hanhua Chen, Hai Jin, Lingchao Guo, “Sink-Free
Audio-on-Demand over Wireless Sensor Networks”,
VOL. 65, NO. 5, MAY 2016
[12] G. WernerAllen, K. Lorincz, J. Johnson, J. Lees, and M.
Welsh,“Fidelity and yield in a volcano monitoring sensor
network,” in Proc. 7th Symp. Oper. Syst. Des.
Implementation, Seattle, WA, USA,Nov. 2006, pp. 381–396.

[13] X. Deng and Y. Yang, “Online adaptive compression in
delay sensitivewireless sensor networks,” IEEE Trans.
Comput., vol. 61,
no. 10, pp. 1429–1442, Oct. 2012.
[14] L. Li, G. Xing, L. Sun, and Y. Liu, “QVS: Quality-aware
voice streaming for wireless sensor networks,” in Proc. Int.
Conf. Distrib. Comput. Syst., Montreal, QC, Canada, June
2009, pp. 450–457.
[15] E. Soroush, K. Wu, and J. Pei, “Fast and quality-
guaranteed data streaming in resource-constrained sensor
networks,” in Proc. MobiHoc, Hong Kong, China, May 2008,
pp. 391–400.
[16] S. Misra, M. Reisslein, and G. Xue, “A survey of
multimedia streaming in wireless sensor networks,” IEEE
Commun. Surveys Tuts., vol. 10, no. 1-4, pp. 18–39, Fourth
Quarter 2008.
[17] S. Ratnasamy, B. Karp, L. Yin, F. Yu, D. Estrin, R.
Govindan, and S. Shenker, “GHT: A geographic hash table
for data centric storage,” in Proc. 1st ACM Int. Workshop
Wireless Sens. Netw. Appl., Atlanta, GA, USA, Sept. 2002,
pp. 78–87.
[18] B. H. Bloom, “Space/time trade-offs in hash coding
with allowable errors,” Commun. ACM, vol. 13, no. 7, pp.
422–426, 1970.
[19] S. Nath, P. B. Gibbons, S. Seshan, and Z. R. Anderson,
“Synopsis diffusion for robust aggregation in sensor
networks,” in Proc. 2nd Int. Conf. Embedded Netw. Sens.
Syst., 2004, pp. 250–262.
[20] D. Koutsonikolas, S. M. Das, Y. C. Hu, and I.
Stojmenovic, “Hierarchical geographic multicast routing
for wireless sensor networks,” Wireless Netw., vol. 16, no.
2, pp. 449–466, 2010.
[21] B. Jiang, B. Ravindran, and H. Cho, “Probability-based
prediction and sleep scheduling for energy-efficient target
tracking in sensor networks,” IEEE Trans. Mobile Comput.,
vol. 12, no. 4, pp. 735–747, Apr. 2013.

Effective Audio Storage and Retrieval in Infrastructure less Environment over WSN

More Related Content

What's hot (19)

Similar to Effective Audio Storage and Retrieval in Infrastructure less Environment over WSN (20)

More from IRJET Journal (20)

Recently uploaded (20)

Effective Audio Storage and Retrieval in Infrastructure less Environment over WSN