TOWARDS INCREASING THE REUSABILITY OF THE

WIRELESS SENSOR NETWORK PROTOCOLS

Sonia Hashish

Department of Computer Science, College of Computer and Information System,

Um-Alqura University, Makkah, PO Box 715, Saudi Arabia

Keywords: Protocol Reusability, Sensor Networks, Network Infrastructure, Self-organization.

Abstract: Increasing the reusability of the wireless sensor network protocols requires decoupling the underlying

communication primitives from the upper layer protocols primitives. One way to achieving this goal is

unifying the whole software stack architecture. This unifying process would significantly increase the

overhead and affect the resulting performance. It is still unknown whether this huge unifying process will

provide the required benefits to the protocol designers. Building a generic infrastructure at the level of

physical links is a promising step towards increasing the reusability of upper layer protocols. To be

described as a generic, the infrastructure should efficiently support different upper layer protocols and

different communication configurations. It should also provide logical relationships among nodes without

hiding the physical relationships. Moreover, Failures should neither destruct the infrastructure nor hinder the

upper layer operations. Building such infrastructure is very challenging and is still an open research

problem.

1 INTRODUCTION

The random deployment process in wireless Sensor

network WSN results in an arbitrary network graph

that is referred to as a network topology (Akyildiz

2002); communication over such a graph is secured

via multiple hops. This mode of communication

implies that each node is able to play the role of a

router, and thus forward the messages over multiple

hops on behalf of other nodes—the arbitrary

topology graph is structured by the communication

algorithm (Yoneki and Bacon, 2005). The

communication process then takes place through the

primitives of the communication algorithm;

Most of the current proposed protocols for WSN

are described as cross-layered protocols (Wu 2007),

which reflects the fact that these protocols are in

most cases tied up to specific communication

algorithm which is optimized for a specific network

deployment. For example, spanning tree based

algorithms opt for building the tree structure over

the arbitrary graph. Then the communication process

takes place using the tree primitives.

Decoupling the communication protocols from

the application semantics to increase the modularity

and reusability is still an open research direction that

should be carefully studied. Given that the sensor

network is mainly application oriented, the benefits

of the clear cut between various networking aspects

over the current cross- layer approaches require

more research efforts to be clearly identified. To

date, little work has been conducted towards

unifying protocol design.

2 PROTOCOLS REUSABILITY

One of the research concerns that is not fully

explored in the literature is the degree of

independency between various network protocols.

Most of the current proposed protocols merge

functions from different networking levels. To date,

little work has been done towards unifying the

design of WSN protocols. WSN is application-

specific in the first place. It poses many challenges

that motivate the production of hundreds of

protocols at each networking level. The different

hardware characteristics and the different

requirements of the upper layer applications boost

the protocol productivity of WSN (Whitehouse et al.

2004).

Numerous protocols that target aggregation,

Hashish S..

TOWARDS INCREASING THE REUSABILITY OF THE WIRELESS SENSOR NETWORK PROTOCOLS.

DOI: 10.5220/0003904800950098

In Proceedings of the 1st International Conference on Sensor Networks (SENSORNETS-2012), pages 95-98

ISBN: 978-989-8565-01-3

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

routing, dissemination, medium access and topology

control have been developed. The main notice

regarding the wide scope of WSN protocol

development is that the performance of a given

protocol is tied to the underlying assumptions about

the rest of the system. When such assumptions are

varied, degradation in performance is noticed. The

variety of possible assumptions about the system

decreases the reusability of the developed protocols.

A new research direction towards unifying WSN

software architecture that increases the modularity

and reusability of the designed protocols is

established in (Culler e al. 2005;

Cheng et al. 2006).

In (Culler et al. 2005) the authors discuss the

narrow waist architecture where sensor protocol

(SP) resides between the network layer and the data

link layer. They describe the rules by which the

network services could be arranged over the layered

architecture. They also discuss the neighbour’s

management issue.

Leveraging the SNA (Culler et al. 2005), a

modular network-layer for sensor networks that sits

atop SP is proposed (

Cheng et al. 2006). Their main

concern is to ease the implementation of new

protocols by increasing code reuse and runtime

sharing. Code reuse provides a rapid protocol and

application development. On the other hand, run-

time sharing reduces code and resources consumed.

The authors discuss the trade-off between

functionality decomposition and complexity. They

find that finding the right granularity at which to

break up the functionality at the network layer is

challenging. Unnecessary runtime overhead could

result from a very fine-grained decomposition while

a too coarse decomposition reduces the level of

sharing, which in turn increases the

reimplementation.

Gnawali (2006)

proposes Tenet architecture,

which is complementary to the narrow waist

architecture proposed in the work of (Culler et al.

2005). Tenet architecture does not address the

modularity of the software. It restricts the placement

of the application functionality in a multi- tier

system. In Tenant, the sensor level tier can be

implemented on SP. Tenet shares some similarities

with the Internet's end-to-end principle (Saltzer et al.

1984), yet it is based on specific tiered network

technology.

The above solutions target achieving complete

standard software stack architecture for sensor

networks. This is still far away from being a reality.

There is a conviction in the literature that this

unifying process would significantly increase the

overhead and affect the resulting performance (Ali

and Langendoen, 2007). It is still unknown whether

this huge unifying process will provide the required

benefits to the protocol designers.

Building a generic infrastructure at the level of

physical links is a promising step towards increasing

the reusability of upper layer protocols. To build

such infrastructure over sensor networks, the

literature explores two approaches; the first is to

construct an in-line infrastructure that supports a

specific process, such as routing or data aggregation.

This model is usually optimized to efficiently

achieve an upper goal such as minimizing

congestion. Clustering and tree-based approaches

are the most utilized techniques to build such

infrastructure. They provide nodes with the means to

self-organize and thus achieve unstructured overlays

(Younis et al. 2006). Operations over such overlays

are usually based on flooding mechanisms (Olariu et

al. 2004); failure handling and maintenance require

cascade updates throughout the network. In addition,

the resulting infrastructure cannot be utilized by

different protocols.

The second approach is building an infrastructure

that is not tied to any upper protocol. This is a

general purpose infrastructure, which should be able

to support different upper layer processes with equal

efficiency. To be generic, we claim that the

infrastructure should adhere to the following design

objectives:

(i) Generic: efficiently supporting different

upper layer protocols (e.g., routing, data collection,

data aggregation and broadcasting).

(ii) Flexible: efficiently supporting different

communication configurations (both multi- hop and

data mule- like communication).

(iii) Maintainable: failures neither destruct the

infrastructure nor hinder the upper layer operations.

(iv) Complete: providing logical relationships

among nodes without hiding physical relationships.

Taking into consideration that sensor networks

are application-oriented and have scarce resources,

the problem of building such generic infrastructure

is challenging.

Building generic infrastructure over sensor

networks is studied in (Olariu et al. 2005). The

authors develop a virtual infrastructure in terms of

coronas and wedges. They consider the case of a

static sensor network where all nodes are static. The

sink, named as Training Agent (TA), is assumed to

be at the centre of the network, and it is assumed

that the TA has multiple-levels transmission range.

The TA takes the burden of training the nodes to

acquire knowledge about their position with respect

to it (TA). The position is considered as the (wedge,

SENSORNETS 2012 - International Conference on Sensor Networks

corona) where the node is located; however, the

protocol is centralized and is based on global

information. The number of coronas to be created

should be known to the TA before it creates them. In

addition, the mechanism proactively divides the

nodes into subsets and requires synchronization of

the wakeup times of each subset of sensor nodes and

the level of the transmission range of the sink at that

time. This model is extended to account for the TA's

mobility in (Olariu et al. 2007). Mobility is

considered for achieving the QoS requirements of

the applications rather than for infrastructure

organization. Moreover, the authors do not specify

the effects of involving multiple sinks/TAs on the

constructed infrastructure. For example, how to

proactively determine the number of coronas per

each TA is not answered.

A multi-scale communication overlay is

developed in (Palchaudhuri et al. 2005) to support

upper layer protocols. The protocol belongs to the

clustering-based approaches. Nodes are organized

into cells, super-cells and so on. A self-election

mechanism based on sending periodic beacons is

used to form the hierarchical overlay. As in most

clustering based approaches, maintaining the whole

structure requires topological updates to be

broadcasted to all nodes, and re-clustering is

performed to adapt to the changes. This introduces

extra overhead that could participate in draining the

resources of sensor nodes.

Building logical overlays, such as Distributed

Hash Tables DHTs, has long been the focus of

research. Sensor nodes have scarce resources in

terms of energy, bandwidth and communication,

which render DHTs unsuitable; the main reason

being the belief that DHT overlays produce extra

overhead compared to the benefits they provide to

the upper layer applications (Ali and Langendoen,

2007). When mobility is considered, movement of

the nodes may quickly change the topology, thus

resulting in an increase in the overhead messages for

topology maintenance and movement management.

An attempt to fill in the space between the

logical and physical infrastructure is proposed in

(Caesar et al. 2006). Authors proposed a Virtual

Ring Routing (VRR) protocol where logical rings

are constructed over the link layer. The protocol is

inspired by DHT mechanisms, and provides both

point-to-point and DHT like operations. The

protocol creates logical rings that do not keep the

node proximity. Moreover, all nodes within the

network should have unique logical addresses

(identifiers) that are globally ordered. In addition,

the protocol is optimized only for routing processes.

A promising approach for building a generic

infrastructure that adheres to the design objectives

mentioned above has been proposed in (Hashish and

Karmouch, 2009). The authors proposed Layered

Infrastructure Protocol (LIP). LIP exploits mobility

to organize sensor nodes, and form a generic flexible

infrastructure that could be leveraged by upper layer

protocols. LIP allows mobile robots/sinks to

discover physical co centric circular layers within

the arbitrary network topology. LIP creates physical

co centric circular layered infrastructure (CLI)-the

resulting CLI infrastructure guarantees the proximity

of the nodes. Nodes that are neighbours in the

infrastructure are physically neighbours; each layer

in CLI is assigned a mobile robot that acts as a probe

to access the data and monitor the layer. Access

positions are selected dynamically at each layer to

act as anchors for the probes to visit at their

associated layers.

CLI has the ability to trade mobility overhead vs.

communication overhead (higher number of access

points implies that data travels in smaller number of

hops to access points, hence to be offloaded to the

moving robot/sink). Moreover, layers in CLI are

managed separately by the associated mobile robots.

This makes CLI a rich environment for developing

efficient upper layer protocols (Hashish, 2010). It

provides varieties of communication configurations

which support both multi-hop and data-mules

regimes of communication. It also provides a high

degree of reliability to the upper layer applications

while reducing the overall energy consumption. This

ability to cope with failures makes it a good

candidate for sensor networks applications;

Applications based on adaptive allocation of mobile

sinks, applications based on nodes scheduling and

applications based on multi-granularities

communications could be efficiently developed atop

of CLI.

3 CONCLUSIONS

In this paper, we show that decoupling the

communication protocols from the application

semantics to increase the modularity and reusability

is still an open research problem. The main

approaches for increasing the reusability of the

wireless sensor network protocols have been

discussed. Achieving standard software stack

architecture for sensor networks is still far away

from being a reality. Building a generic

infrastructure at the level of physical links is a

promising step toward increasing the reusability of

TOWARDS INCREASING THE REUSABILITY OF THE WIRELESS SENSOR NETWORK PROTOCOLS

upper layer protocols. This is very challengeable in

sensor networks that feature scares resources. Some

of the existing solutions and their limitations have

been described. Limitations of the existing solutions

have been mentioned.

REFERENCES

Abbasi, A., Younis, M., 2007. A survey on clustering

algorithms for wireless sensor networks. Computer

Communications, Volume 30, Issue 14-15 Pages

2826-2841.

Akyildiz, I., et al., 2002. Wireless sensor networks: a

survey. Computer Networks, Vol. 38, pp. 393- 422.

Ali., M., Langendoen, K., 2007. A case for peer-to-peer

network overlays in sensor networks. In International

Workshop on Wireless Sensor Network Architecture

(WWSNA'07).

Caesar, M., Castro, M., Nightingale, E., Rowstron, A.,

2006. Virtual ring routing: network routing inspired by

DHTs. In ACM SIGCOMM.

Cheng, E., Fonseca, R., Kim, S., Moon, D., Tavakoli, A.,

Culler, D., Shenker, S., Stoica, I., 2006. A modular

network layer for sensornets. In OSDI’06, 6-8

November Seattle, WA, USA, 249-262.

Culler, D., Dutta, P., Cheng, E., Fonseca, R., Hui, J.,

Levis, P., Polastre, J., Shenker, S., Stoica, I., Tolle, G.,

Zhao. J., 2005. Towards a sensor net architecture:

Lowering the waistline. In HotOS, Santa Fe, NM, pp.

24 – 24.

Gnawali, O., Greenstein, B., Jang, K-Y., Joki, A., Paek, J.,

Vieira, M., Estrin, D., Govindan, R., Ko, E., 2006. The

Tenet architecture for tiered sensor networks. In ACM

Sensys, November 2006 Boulder, Colorado, USA,

153-166.

Hashish, S., Karmouch, A., 2009. Mobility-based generic

infrastructure for large scale sensor network

architecture. In IEEE ICC ‘09, 13-18 June Dresden,

Germany.

Hashish S., 2010. Efficient communication protocols for

sensor network architecture with multiple mobile

sinks. Thesis, (PhD). University of Ottawa, Canada.

Olariu, S., Wadaa, A., L. Wilson, L., Eltoweissy, M.,

2004. Wireless sensor networks - Leveraging the

virtual infrastructure. IEEE Network, Volume 18,

July/August 2004, 51-56.

Olariu, S., Wadaa, A., Xu Q., Stojmenovic, I., 2005. A

virtual infrastructure for wireless sensor networks. In:

Stojmenovic, I., ed., Handbook of Sensor Networks:

Algorithms and Architectures, Wiley, 170-140.

Olariu, S., Eltoweissy, M., Younis, M., 2007. ANSWER:

AutoNomouS netWorked sEnsoR system. Journal of

Parallel and Distributed Computing, 67(1):111-124.

Palchaudhuri, S., Kumar, R., Baraniuk, R., Johnson, D.,

2005. Design of adaptive overlays for multi-scale

communication in sensor network. In Proceedings of

the International Conference on Distributed Compu-

ting in Sensor Systems (DCOSS). 2005, 173-190.

Saltzer, J., Reed, D., Clark, D., 1984. End-to-end

arguments in system design. ACM Transactions on

Computer Systems, 2(4):277.

Wu, J., 2007 Reliable routing protocols for dynamic

wireless ad hoc and sensor networks. Thesis, (PhD).

University of Twente, Enschede, the Netherlands.

Whitehouse, K., Sharp, C., Brewer, E., Culler, D., 2004.

Hood: a neighborhood abstraction for sensor networks.

In: MobiSYS '04: Proceedings of the 2nd international

conference on Mobile systems, applications, and

services, ACM Press (2004) 99-110.

Yoneki, E., Bacon, J., 2005. Mobility-Based

Communication in Wireless Sensor Networks, A

survey of Wireless Sensor Network technologies:

research trends and middleware’s role. Technical

Report: University of Cambridge, UCAM-CL-TR-646

Younis, O., Krunz, M., Ramasubramanian, S., 2006. Node

clustering in wireless sensor networks: recent

developments and deployment challenges. IEEE

Network, Volume 20, Issue 3. DOI: 10.1109.

SENSORNETS 2012 - International Conference on Sensor Networks