Chapter 1: (Data-Driven Wireless Networks: A Perspective)

Author(s):
M. Ateeq (The Islamia Univ of Bahawalpur, muhammad.ateeq@iub.edu.pk)
*M.K. Afzal (COMSATS Univ Islamabad, Wah Campus, khalilafzal@ciitwah.edu.pk)

1.1 Evolution of Wireless Networks

Wireless has become a prevalent medium for modern communication systems. At its very

core, we are realizing the internet of things that uses all sorts of communication networks

including WiFi, wireless sensor networks, and cellular networks. The physical environments

around us have rich sources of information that we sense and appropriately react to. With

ever-growing needs of the world and advancement in technology, environments are intended

to be smart where both sensing and decision making are autonomous, fast, and robust. In

order to facilitate such ubiquitous sensing and put the sensed data to use, there is an

established need for state-of-the-art resilient and adaptive communication infrastructure. To

serve this purpose, Wireless Sensor Networks (WSNs) [1] have been around for the past two

decades. WSNs consist of resource-constrained nodes capable of sensing and communicating

over low power radio interfaces. The number of nodes may range from 10's to 1000's,

depending on the application domain and deployment scenario. These nodes collect large-

scale data and relay it to a base station that carries out the relevant processing tasks [2].

To be specific, we shall refer more to WSNs as an example of wireless networks through this

chapter.

1.1.1 WSN Applications

WSNs target broad applications, primarily categorized into monitoring and tracking [3].

Some important and popular domains of interest include environment, industry and

agriculture, infrastructure, military, and body area, etc. In the context of Industrial WSNs

(IWSNs), International Society of Automation (ISA) distributes industrial systems into the

following six classes [4]:

 Closed-loop regulatory control systems,

 Close loop supervisory systems,

 Open-loop control systems,

 Alerting systems, and

 Information gathering systems.

Some particular IWSN application areas include area monitoring, machine health monitoring,

structural monitoring, air/water quality, and waste monitoring [5]. With growing automation,

applications tend to surpass from legacy monitoring systems to more intelligent applications

that have the capability to react to a situation based on the data collected. Moreover, these

applications are expected to learn and adapt their behavior over time as circumstances vary.

Sensor and actor-networks demonstrate this behavior in the form of emergency response,

disaster prevention, industrial automation, and feedback control, etc.

1.1.2 WSNs and IoT

Historically, deployments of WSNs have been domain-specific with tailor-made solution to

serve a particular application scenario. Recently, as the capabilities (e.g., sensing, processing,

communication, and energy replenishment, etc.) of the sensor nodes increase at scale, these

are becoming more and more connected, integrating to realize the Internet of Things (IoT).

Smart homes, buildings, and cities are among the common examples highlighting this

integration of diverse WSN deployments where appliance monitoring, surveillance, water

management, Heating, Ventilation and Air Conditioning (HVAC), and elevator control are

among the applications integrating to form the IoT [6].

1.2 Adaptivity and QoS Challenge

Because the sensor nodes are battery-powered, and often remotely deployed (making it

difficult to replace or replenish), energy consumption has been at the heart of WSN design
since its inception [7]. However, the growth in deployment domains and the diversity of

applications put stringent and assorted Quality of Service (QoS) requirements on WSNs [5].

In addition to energy, factors like reliability, delay, throughput, and signal strength have

become prominent overtime [8][9].

1.2.1 Legacy Solutions

To address QoS challenge, most of the research has focused devising new Medium Access

Control (MAC) [9] or routing protocols [10], and cross-layer approaches [11]. Optimizing

any QoS metric often requires solving an optimization problem that are intractable and

regarded as NP-hard, in general. Moreover, confronting applications where multiple QoS

metrics have to be considered simultaneously, facilitating QoS becomes a Multi-Objective

Optimization (MOO) problem, adding adversely to the computational complexity [12]. The

solutions for MOO problems are often based on approximations, trading-off accuracy for the

sake of complexity. Further, the legacy approaches lack in yielding in an explication that can

facilitate adaptation as the network evolves and changes occur.

1.2.2 Data-Driven Solutions

In order to devise a solution accommodating multiple QoS metrics with sufficient accuracy,

and inculcating adaptivity as the network undergoes changes, data-driven approaches have

become popular. Such systems use performance data coupled with intelligent algorithms

based on Artificial Intelligence (AI) and Machine Learning (ML) to accurately predict the

QoS metrics [13]. Intelligence, adaptivity, affordable complexity, and availability of

performance data are among the prominent factors driving the adoption of data-driven

techniques for QoS predictions in WSNs.

1.2.3 Aim of This Book

The aim of this book is to introduce data-driven approached applied to solving various

problems in wireless networks. These problems include QoS (as discussed in this
introductory chapter), security solutions, ………….

1.3 The Data-Driven Paradigm

Intelligence and adaptivity are prominent features of modern computing systems, driven by

AI and ML. The same paradigm shift is occurring in the communication domain. Data-

Driven Networking (DDN) has been proposed with centralized control [14]. Using the data-

driven approach, WSNs can learn from their own on-going performance recordings and adapt

the behavior in accordance with the in-situ QoS requirements.

The chief benefit of using the centralized approach enables running the sophisticated state-of-

the-art learning algorithms on a substantial volume of data without putting load on the

constrained sensor nodes. A typical flow of the potential data-driven paradigm for WSNs is

shown in Figure 1.1. The performance data from heterogeneous deployments of WSNs is

collected and centrally processed using techniques like ML to learn the QoS attributes based

on various characteristics and parameters of the network. The prediction results are used for

making intelligent decisions to meet desired QoS goals. This learning loop continues, as the

transitions in different settings of the WSNs occur, and system evolves.

1.3.1 QoS Metrics and Adaptivity

In wireless systems, QoS is primarily seen in the facets of timeliness, bandwidth requirement,

and reliability. In addition to these, no QoS system can be realized without earnest

consideration for energy in case of WSNs [15]. Moreover, adaptivity is a characteristic of

increasing importance in modern communication systems. QoS metrics, their relationships

combined with an adaptivity loop are shown in Figure 1.2. In the following, these QoS

factors are briefly explained.

Timeliness

Timeliness is measured primarily using the delay as a metric. Depending on the particular

context under consideration, the definitions of delay can vary slightly.

In the norm of this thesis, the delay is defined as the time between the first transmission from

sender and reception at receiver including retransmission, if any. Delay is an important

metrics for many application domains where certain time-constraints apply in decision-

making [16]. Breach of time-limit can result in different kinds of losses including safety and

finance. Such cases occur in health-care, IWSNs, and any emergency response situations.

Other metrics and terms frequently used to measure timeliness include latency, response time,

and jitter, etc.

Reliability
Reliability is of paramount importance in the performance of any wireless communication

system because of inherent vulnerability to interference, and error-prone links. Moreover,

standards like IEEE 802.15.4 use Industrial, Scientific, and Medical (ISM) band for low-

power communication. Since other standards, such as WiFi use the same bands with a

significantly higher power, achieving reliability can be challenging and power consumption

becomes a major concern as well. At the Physical (PHY) layer, Bit Error Rate (BER) is used

as an indicator of reliability. However, at higher layers, packet delivery ratio is used since

data is grouped in the form of packets.

Packet delivery ratio is important for effective monitoring and response. The decrease in

packet delivery ratio does not just downgrade the performance of WSN but also causes a hike

in energy consumption due to retransmissions. Coupled with delay, packet delivery ratio

defines an important class of applications, regarded as time- and mission-critical. In such

settings, data must be reliably delivered observing certain time-limits to avoid the catastrophe

of some kind [17].

Bandwidth

WSNs typically operate with limited bandwidth. IEEE 802.15.4, a popular low-power

communication standard has a maximum data rate of 250Kbps [18]. Intrinsic limits of

reliability make it even more difficult to achieve good throughput. Historically, WSNs were

not conceived to host data-intensive application having high throughput demands. However,

modern applications with multimedia contents bring forth the importance of effective

utilization of bandwidth in order to facilitate critical operations like remote patient

monitoring [19]. Therefore, bandwidth requirement and its effective utilization coupled with

other metrics like delay and reliability become crucial in WSN design and associated

performance objectives [20].

Energy

Energy has been a prevalent consideration in WSN design since its inception. The primary
reason is that the sensor nodes are generally battery-powered and often so remotely deployed

that repeated access to the nodes is not possible. Therefore, energy consumption has to be

considered alongside other QoS goals. Some of the metrics associated with energy include

lifetime maximization, transmission power control, and energy consumption minimization [7]

[21]. In this thesis, I have used energy consumed per useful bit delivered as the metric to

predict.

Adaptivity

Adaptivity is a trait that all modern computer systems are expected to adapt as they undergo

changes and evolution. Although the term "adaptive" has frequently been used for different

research dimensions in the domain of WSNs; however, the focus on the real definition of an

adaptive QoS system has lacked a great deal [11][22][23]. To be truly adaptive, a

communications system must be able to inculcate changes automatically to facilitate QoS

where inputs and other diverse characteristics of the systems change/evolve over time. Such a

behavior is also regarded as self-organizing and self-optimizing. Thus adaptivity operates in a

loop that continues to collect and observe real-time data, intelligently learn to influential and

conclusive relationships, and patterns, and hence provide decision optimizing support to meet

desired QoS objectives [14].

1.4 Outline of the Book

……

References

[1] Akyildiz, Ian F., Weilian Su, Yogesh Sankarasubramaniam, and Erdal Cayirci. "Wireless sensor
networks: a survey." Computer networks 38, no. 4 (2002): 393-422.
[2] Rawat, Priyanka, Kamal Deep Singh, Hakima Chaouchi, and Jean Marie Bonnin. "Wireless sensor
networks: a survey on recent developments and potential synergies." The Journal of
supercomputing 68, no. 1 (2014): 1-48.
[3] Arampatzis, Th, John Lygeros, and Stamatis Manesis. "A survey of applications of wireless
sensors and wireless sensor networks." In Proceedings of the 2005 IEEE International Symposium on,
Mediterrean Conference on Control and Automation Intelligent Control, 2005., pp. 719-724. IEEE,
2005.
[4] Zand, Pouria, Supriyo Chatterjea, Kallol Das, and Paul Havinga. "Wireless industrial monitoring
and control networks: The journey so far and the road ahead." Journal of sensor and actuator
networks 1, no. 2 (2012): 123-152.
[5] Raza, Mohsin, Nauman Aslam, Hoa Le-Minh, Sajjad Hussain, Yue Cao, and Noor Muhammad
Khan. "A critical analysis of research potential, challenges, and future directives in industrial wireless
sensor networks." IEEE Communications Surveys & Tutorials 20, no. 1 (2017): 39-95.
[6] Gubbi, Jayavardhana, Rajkumar Buyya, Slaven Marusic, and Marimuthu Palaniswami. "Internet
of Things (IoT): A vision, architectural elements, and future directions." Future generation computer
systems 29, no. 7 (2013): 1645-1660.
[7] Rault, Tifenn, Abdelmadjid Bouabdallah, and Yacine Challal. "Energy efficiency in wireless
sensor networks: A top-down survey." Computer networks 67 (2014): 104-122.
[8] Chen, Dazhi, and Pramod K. Varshney. "QoS support in wireless sensor networks: a survey."
In International conference on wireless networks, vol. 233, pp. 1-7. 2004.
[9] Yigitel, M. Aykut, Ozlem Durmaz Incel, and Cem Ersoy. "QoS-aware MAC protocols for wireless
sensor networks: A survey." Computer Networks 55, no. 8 (2011): 1982-2004.
[10] Yetgin, Halil, Kent Tsz Kan Cheung, Mohammed El-Hajjar, and Lajos Hanzo Hanzo. "A survey
of network lifetime maximization techniques in wireless sensor networks." IEEE Communications
Surveys & Tutorials 19, no. 2 (2017): 828-854.
[11] Al-Anbagi, Irfan, Melike Erol-Kantarci, and Hussein T. Mouftah. "A survey on cross-layer
quality-of-service approaches in WSNs for delay and reliability-aware applications." IEEE
Communications Surveys & Tutorials 18, no. 1 (2014): 525-552.
[12] Fei, Zesong, Bin Li, Shaoshi Yang, Chengwen Xing, Hongbin Chen, and Lajos Hanzo. "A survey
of multi-objective optimization in wireless sensor networks: Metrics, algorithms, and open
problems." IEEE Communications Surveys & Tutorials 19, no. 1 (2016): 550-586.
[13] Jiang, Junchen. "Enabling Data-Driven Optimization of Quality of Experience in Internet
Applications." PhD diss., Intel Corporation, 2017.
[14] Jiang, Junchen, Vyas Sekar, Ion Stoica, and Hui Zhang. "Unleashing the potential of data-driven
networking." In International Conference on Communication Systems and Networks, pp. 110-126.
Springer, Cham, 2017.
[15] Chen, Ray, Anh Phan Speer, and Mohamed Eltoweissy. "Adaptive fault-tolerant QoS control
algorithms for maximizing system lifetime of query-based wireless sensor networks." IEEE
Transactions on dependable and secure computing 8, no. 2 (2010): 161-176.
[16] Kim, Joohwan, Xiaojun Lin, Ness B. Shroff, and Prasun Sinha. "Minimizing delay and
maximizing lifetime for wireless sensor networks with anycast." IEEE/ACM Transactions on
networking 18, no. 2 (2009): 515-528.
[17] Mahmood, Muhammad Adeel, Winston KG Seah, and Ian Welch. "Reliability in wireless sensor
networks: A survey and challenges ahead." Computer networks 79 (2015): 166-187.
[18] Lu, Gang, Bhaskar Krishnamachari, and Cauligi S. Raghavendra. "Performance evaluation of the
IEEE 802.15. 4 MAC for low-rate low-power wireless networks." In IEEE International Conference
on Performance, Computing, and Communications, 2004, pp. 701-706. IEEE, 2004.
[19] Tung, Hoi Yan, Kim Fung Tsang, Hoi Ching Tung, Kwok Tai Chui, and Hao Ran Chi. "The
design of dual radio ZigBee homecare gateway for remote patient monitoring." IEEE Transactions on
Consumer Electronics 59, no. 4 (2013): 756-764.
[20] Jabbar, Sohail, Abid Ali Minhas, Muhammad Imran, Shehzad Khalid, and Kashif Saleem.
"Energy efficient strategy for throughput improvement in wireless sensor networks." Sensors 15, no. 2
(2015): 2473-2495.
[21] Ren, Ju, Yaoxue Zhang, Kuan Zhang, Anfeng Liu, Jianer Chen, and Xuemin Sherman Shen.
"Lifetime and energy hole evolution analysis in data-gathering wireless sensor networks." IEEE
transactions on industrial informatics 12, no. 2 (2015): 788-800.
[22] Ezdiani, Syarifah, Indrajit S. Acharyya, Sivaramakrishnan Sivakumar, and Adnan Al-Anbuky.
"An IoT environment for WSN adaptive QoS." In 2015 IEEE International Conference on Data
Science and Data Intensive Systems, pp. 586-593. IEEE, 2015.
[23] Ezdiani, Syarifah, Indrajit S. Acharyya, Sivaramakrishnan Sivakumar, and Adnan Al-Anbuky.
"Wireless sensor network softwarization: Towards WSN adaptive QoS." IEEE Internet of Things
Journal 4, no. 5 (2017): 1517-1527.