Data Collection Protocols in Wireless Sensor Networks

*Koppala Guravaiah, Arumugam Kavitha and Rengaraj Leela Velusamy*

#### **Abstract**

In recent years, wireless sensor networks have became the effective solutions for a wide range of IoT applications. The major task of this network is data collection, which is the process of sensing the environment, collecting relevant data, and sending them to the server or BS. In this chapter, classification of data collection protocols are presented with the help of different parameters such as network lifetime, energy, fault tolerance, and latency. To achieve these parameters, different techniques such as multi-hop, clustering, duty cycling, network coding, aggregation, sink mobility, directional antennas, and cross-layer solutions have been analyzed. The drawbacks of these techniques are discussed. Finally, the future work for routing protocols in wireless sensor networks is discussed.

**Keywords:** wireless sensor networks, routing protocols, data collection, network lifetime, energy efficiency, fault tolerance, low latency

#### **1. Introduction**

Wireless sensor networks (WSNs) [1] are distributed among environment with lightweight and small sensor nodes. These sensor nodes are used to measure the parameters of environment. Some of such parameters are vibration, pressure, sound, movements, temperature, humidity, etc. The sensors are well coordinated and connected to the base station (BS) or sink using wireless communication for forwarding sensed information. Due to this, many IoT-based applications such as home applications [2], vehicular monitoring [3], medical applications, structural monitoring, habitat monitoring, intrusion detection, tracking for military purpose, etc., are using WSNs for data collection [1, 4, 5].

Ad hoc and cellular network routing protocols are not suitable for sensor networks due to the sensor node design challenges such as node deployment, node mobility, and limited resource constraints (battery, communication, and processing capabilities) [6]. In WSNs, large number of sensor nodes are deployed for specific application due to this global addressing which is too difficult to maintain. Due to this large number, nodes located in the same area may generate redundant data and transmit to BS. This leads to bandwidth wastage and network traffic which in turn effects the more energy consumption. Another main resource constraint of a sensor node is limited battery power due to battery replacement or recharge not being possible in most of the WSN applications. WSN has a wireless communication medium, which


Transmitting of sensed data to BS is either by data dissemination (data diffusion) or data gathering (data delivery) [8]. Data/queries (network setup/management and/ or control collection commands) propagation throughout the network is done in the data dissemination stage. Low latency is the main issue for disseminating data/

Data delivery or data gathering is the forwarding of sensed data to the BS. The main aim of data gathering is to maximize the number of rounds of data transferring toward BS before the network died. This will be achieved by minimizing

Single-hop or multi-hop is the basic communication technique between source sensor node and BS in data gathering. Sensed data are forwarded directly to BS in the single-hop communication. In multi-hop [9], the sensed data are forwarded to the base station with the help of intermediate sensor nodes. In multi-hop routing, energy conservation, route discovery, QoS, and low latency are the major issues. Introducing mobility in sink nodes, called mobile sinks or mobile collectors [10] is also a single-hop communication. In this network, mobile sink nodes move along a trajectory path to access the data from all source sensor nodes in a single-hop fashion. The trajectory path identification is the important step in this single-hop communication to cover all the nodes throughout the network. Energy conservation and mobility are the major issues in mobility-based single-hop data transmission.

Different classification of data collection routing protocols [6, 11–15] are proposed in recent years by researchers. **Figure 1** shows the different classifications of

Network architecture-based classification was presented by Akkaya et al. [6] in 2005. According to Akkaya et al., routing protocols are classified as data-centric, hierarchical, and location-based protocols. Sink disseminating the queries in network to get the sensor data from sensor nodes is the work of data-centric protocols. In cluster- or hierarchical-based protocols, network of nodes is divided into clusters and each cluster is managed by the cluster head (CH). Each CH will receive the sensed data from the corresponding cluster member and forward it to the BS. Aggregation techniques can be used by the CH to save energy while forwarding to BS. Geographic- or location-based protocols are considering the position informa-

Multipath, query-based, negotiation-based, quality of service (QoS)-based, and

coherent-based protocols are the classification of routing protocols as given by Karaki et al. [11]. In multipath routing, multiple paths are selected for achieving a variety of benefits such as reliability, fault tolerance, and increased bandwidth. Data

energy consumption and delay for each transmission.

*Data Collection Protocols in Wireless Sensor Networks DOI: http://dx.doi.org/10.5772/intechopen.93659*

**2.1 Taxonomy of data collection protocols**

data collection routing protocols.

tion of sensor nodes for routing.

*Taxonomy of data collection protocols.*

**Figure 1.**

**85**

queries to BS.

*EE: energy efficiency; LT: lifetime; LL: low latency; FT: fault tolerance; S: scalability; Q: quality of service; R: reliability; L: low; M: medium; H: high.*

#### **Table 1.**

*WSN applications based on data collection requirements.*

leads to an increased probability of collisions in the data communication process and which impacts on the network performance. While designing a new data collection routing protocol and achieving its requirements such as coverage area, data accuracy, and low latency, we need to consider the above stated issues [7].

In WSN, collection of sensed data can be done in a regular or non-regular mode. Data have to be collected continuously from sensor nodes in regular mode. Whereas, in the non-regular mode, the data have to be collected at some periodic intervals from sensor nodes. **Table 1** refers to different design metrics such as energy efficiency (EE), lifetime (LT), low latency (LL), fault tolerance (FT), security (S), quality of service (Q), and reliability (R), which are considered with the level of importance [low (L), medium (M), and high (H)] for different WSN applications.

This chapter's main objective is the better understanding of data collection protocol with respect to network lifetime, energy conservation, fault tolerance, and low latency. In addition to this, understanding of some existing techniques such as multi-hop, clustering, duty cycling, aggregation, directional antennas, network coding, sink mobility, and cross-layer solutions for achieving these parameters.

#### **2. Data collection**

For sensing the data from the environment and transferring to the BS, the sensor nodes are deployed at specific locations. The data collection's main goal is accuracy of sensing and transmitting the data to BS without any information loss and delay.

#### *Data Collection Protocols in Wireless Sensor Networks DOI: http://dx.doi.org/10.5772/intechopen.93659*

Transmitting of sensed data to BS is either by data dissemination (data diffusion) or data gathering (data delivery) [8]. Data/queries (network setup/management and/ or control collection commands) propagation throughout the network is done in the data dissemination stage. Low latency is the main issue for disseminating data/ queries to BS.

Data delivery or data gathering is the forwarding of sensed data to the BS. The main aim of data gathering is to maximize the number of rounds of data transferring toward BS before the network died. This will be achieved by minimizing energy consumption and delay for each transmission.

Single-hop or multi-hop is the basic communication technique between source sensor node and BS in data gathering. Sensed data are forwarded directly to BS in the single-hop communication. In multi-hop [9], the sensed data are forwarded to the base station with the help of intermediate sensor nodes. In multi-hop routing, energy conservation, route discovery, QoS, and low latency are the major issues. Introducing mobility in sink nodes, called mobile sinks or mobile collectors [10] is also a single-hop communication. In this network, mobile sink nodes move along a trajectory path to access the data from all source sensor nodes in a single-hop fashion. The trajectory path identification is the important step in this single-hop communication to cover all the nodes throughout the network. Energy conservation and mobility are the major issues in mobility-based single-hop data transmission.

#### **2.1 Taxonomy of data collection protocols**

Different classification of data collection routing protocols [6, 11–15] are proposed in recent years by researchers. **Figure 1** shows the different classifications of data collection routing protocols.

Network architecture-based classification was presented by Akkaya et al. [6] in 2005. According to Akkaya et al., routing protocols are classified as data-centric, hierarchical, and location-based protocols. Sink disseminating the queries in network to get the sensor data from sensor nodes is the work of data-centric protocols. In cluster- or hierarchical-based protocols, network of nodes is divided into clusters and each cluster is managed by the cluster head (CH). Each CH will receive the sensed data from the corresponding cluster member and forward it to the BS. Aggregation techniques can be used by the CH to save energy while forwarding to BS. Geographic- or location-based protocols are considering the position information of sensor nodes for routing.

Multipath, query-based, negotiation-based, quality of service (QoS)-based, and coherent-based protocols are the classification of routing protocols as given by Karaki et al. [11]. In multipath routing, multiple paths are selected for achieving a variety of benefits such as reliability, fault tolerance, and increased bandwidth. Data

**Figure 1.**

*Taxonomy of data collection protocols.*

leads to an increased probability of collisions in the data communication process and which impacts on the network performance. While designing a new data collection routing protocol and achieving its requirements such as coverage area, data accuracy,

**Data collection Applications EE LT LL FT S Q R** Regular data collection Health care Patient monitoring M M H H H H H

*Wireless Sensor Networks - Design, Deployment and Applications*

Traffic control and monitoring

Environment control in

buildings

Environmental Vehicle tracking and detection

*EE: energy efficiency; LT: lifetime; LL: low latency; FT: fault tolerance; S: scalability; Q: quality of service; R:*

Military Battlefield surveillance H H H H H H H

Public Factory monitoring M M H M M M H Industrial Machine monitoring M M H M L M H Safety Chemical monitoring M M H M M M H Environmental Disaster monitoring H H H H L M M

Agriculture Precision agriculture H H L M L L H

Industrial Managing inventory control M M M L L L M Home Smart home automation M M L L L L M

Animal monitoring H H L L L L M

Disaster damage assessment M M L L L M M

Structural monitoring H H H H M M H

M M H H MHM

MMML LLM

H H L L L MM

In WSN, collection of sensed data can be done in a regular or non-regular mode.

Data have to be collected continuously from sensor nodes in regular mode. Whereas, in the non-regular mode, the data have to be collected at some periodic intervals from sensor nodes. **Table 1** refers to different design metrics such as energy efficiency (EE), lifetime (LT), low latency (LL), fault tolerance (FT), security (S), quality of service (Q), and reliability (R), which are considered with the level of importance [low (L), medium (M), and high (H)] for different WSN

This chapter's main objective is the better understanding of data collection protocol with respect to network lifetime, energy conservation, fault tolerance, and low latency. In addition to this, understanding of some existing techniques such as multi-hop, clustering, duty cycling, aggregation, directional antennas, network coding, sink mobility, and cross-layer solutions for achieving these parameters.

For sensing the data from the environment and transferring to the BS, the sensor nodes are deployed at specific locations. The data collection's main goal is accuracy of sensing and transmitting the data to BS without any information loss and delay.

and low latency, we need to consider the above stated issues [7].

applications.

**Table 1.**

Non-regular data collection

*reliability; L: low; M: medium; H: high.*

*WSN applications based on data collection requirements.*

**2. Data collection**

**84**

acquisition is done by the sink node with the help of query dissemination in querybased routing. All sensor nodes are going to store the data based on the interest of nodes. Then the data are forwarded to the destination only if the sensed or received node data match with the received queries. Data descriptors are used by negotiationbased protocols for reducing redundant data relays through negotiation. QoS-based protocols mainly consider QoS metrics such as delay, throughput, bandwidth, etc., when routing the data to the base station. In coherent routing, the sensed data is transferred directly to the aggregate node. Whereas in noncoherent routing, node data processing is done locally and then is transferred to neighbor nodes. In addition, routing protocols are classified into proactive, reactive, and hybrid protocols depending on path establishment between the source and destination.

clustering, aggregation, network coding, duty cycling, directional antennas, sink mobility, and cross-layer solutions which are used to achieve efficient data

Managing energy of the sensor nodes is the primary concern in WSN because it is the critical constraint of the sensor nodes. Saving of the node energy increases the network lifetime. Sensor node depletes much energy in two significant operations such as environment sensing and communicating sensed data to the BS. Energy consumption is stable for sensing operation because it depends on the sampling rate and does not depend on the other factors such as the topology of network or the location of the sensors. While, data forwarding process depends on them. Hence, energy conservation is feasible by designing an effective data forwarding process. Network lifetime [21] is defined as the period from the starting of the WSN operation to the time when any or a given percentage of sensor nodes die. Hence, the major objective of the data collection protocol is to gather the data with the maximum number of rounds within the lifetime of the network. The data gathering is the vital factor which considers energy saving as well as lifetime. In literature [4, 22], the authors have presented energy-efficient techniques for data collection. Rault et al. [4] have reviewed the energy-saving techniques and its classification such as radio optimization, data reduction, sleep/wake-up schemes, energyefficient routing, and battery repletion. Anastasi et al. [22] in 2009 discussed directions for energy conservation in WSNs and presented the taxonomy of energy conservation techniques such as duty cycling, data driven, and mobility-based

Latency is the period from the time unit that the data generation at the sensor node started to the time unit that data reception was completed at the base station. It is one of the main concerns for time significant applications such as military and medical health-care monitoring. Attaining low latency is a vital concern because of

1.Due to limited constraints of sensor nodes which are more prone to failure.

2.Collisions and network traffic will be increased due to the broadcast nature of

3. Same kind of data will be sensed by densely deployed sensors and transfer to BS will increase the network traffic and exhaust the communication

To deal with the above issues, there is a need for low-latency protocols. Literature [23, 24] presents recent survey works on low-latency routing protocols. Srivathsan and Iyengar [23] have reviewed some key mechanisms to reduce the latency in single-hop and multi-hop wireless sensor networks; such mechanisms are sampling time, propagation time, processing time, scheduling, use of directional antennas, MAC protocols, sleep/wake-up cycles, predictions, use of dual-frequency radios, etc.

collection routing protocols are also presented.

*Data Collection Protocols in Wireless Sensor Networks DOI: http://dx.doi.org/10.5772/intechopen.93659*

**3.1 Design issues in data collection**

*3.1.1 Energy and lifetime*

routing.

*3.1.2 Latency*

the following reasons:

radio channel.

bandwidth.

**87**

Continuous, event-driven, observer-initiated, and hybrid-based on application interest are the different classifications given by Tilak et al. [12] in 2002. The sensor nodes transfer their sensed data at a prespecified rate to the server in the continuous model. Only when an event occurs, the sensor nodes forward data to base station in the event-driven data model. In the observer-initiated model, the observer will give an explicit request, then only the corresponding sensor nodes respond with the results. The combination of above three approaches will be called as hybrid protocols.

Based on data communication functionalities of routing protocols, Kai Han et al. [31], in 2013, classified the routing protocols into unicast, anycast, broadcast, multicast, and converge-cast. One-to-one association between sensor nodes is used in unicast routing. For forwarding the sensed data, unicast routing is using one neighboring node as a relay node. In anycast routing, nodes transfer the sensed data to a potential receiver node of a group. Multicast routing is transferring the data to a selected number of neighbor nodes simultaneously in a single transmission. Broadcast routing uses a one-to-many association; in a single transmission, sensor nodes transfer the data to their all neighbor nodes simultaneously. The data are aggregated at relay nodes and forwarded toward the base station in the converge-cast mechanism. Information exchanges will be done between the pair of sensor nodes in unicast/ anycast. Whereas, multicast/broadcast is required for disseminating commands to sensor nodes, and converge-cast uses to collect the data from sensor nodes.

Routing protocols are classified as classical and swarm intelligence-based protocols by A.M. Zungeru et al. [14]. Further, each protocol is categorized into data-centric, hierarchical, location-based, network flow, and quality of service (QoS) awareness. In addition, they divided the routing protocols into proactive, reactive, and hybrid, depending on the path establishment between the source and destination.

The energy-efficient routing protocols are classified into network structure, communication model, topology-based, and reliable routing, as presented by Pantazis et al. [15]. Network structure routing protocols are classified into flat and hierarchical protocols. Communication model routing protocols can be divided into coherent or query-based and negotiation-based or noncoherent-based protocols. Mobile agent-based or location-based routing protocols are under the category of topology-based routing protocols. Reliable routing protocols are classified as multipath-based or QoS-based.

In addition to the above, some other literature [16–20] also presented different classifications of routing protocol. However, **Figure 1** represents the overall classification of routing protocols in WSN.

#### **3. Major design issues and techniques for data collection**

In this section, some common design issues for data collection, such as energy, lifetime, latency, and fault tolerance are discussed. The techniques such as

clustering, aggregation, network coding, duty cycling, directional antennas, sink mobility, and cross-layer solutions which are used to achieve efficient data collection routing protocols are also presented.
