A seismic survey is a method of obtaining the graphical representation of the earth’s subsurface structure by analysis of seismic waves. It has a vast area of applications, such as volcanic monitoring [ 1 ], earthquake early warning system [ 2 ], landslide monitoring [ 3 ], mineral resources survey [ 4 ], imaging of glaciers and ice sheets to monitor how change in climate affects the subglacial environment [ 5 6 ], etc. It is often achieved using a technique called “Seismic Reflection”. For almost a century, exploration companies commonly employ this method to determine ideal places to drill for oil and gas, as well as monitor and plan enhanced recovery programs [ 7 ]. A seismic survey can either be onshore or offshore. In an onshore survey, a network of sensors called “geophones” are deployed on the survey area, typically in a line or in a rectangular grid or other geometry, according to pre-defined survey parameters. The traditional approach of deploying these networks is to connect each sensor via cable. Cable-based surveys are known to have a lot of disadvantages, such as excess weight, reliability issues, complexities in deployment and maintenance, human resource costs [ 8 ], and other operational costs. According to [ 9 ], cables account for up to 50% of operational costs and 75% of the total equipment weight in land seismic surveys. Moreover, cables are prone to damages by natural and cultural causes, resulting in survey down-time as considerable field time is expended troubleshooting faulty cables [ 10 ]. The growing demand for better quality, high-density subsurface imaging is impelling seismic exploration companies to provide more recording channels in surveys. Future seismic surveys might require recording channels ranging from hundreds of thousands and above [ 11 ]. This will result to even more complex, cumbersome, and expensive logistics around cable-based exploration. To address the problems associated with cables, wireless seismic data acquisition systems have been proposed in recent years. The system utilizes seismic sensor nodes equipped with wireless transceivers to form a network of wireless geophone sensors that employ radio frequency communication technology. Although proposals for wireless acquisition systems date back to few decades ago [ 12 ], recent advancement in wireless technology may pave the way for a complete cable-free wireless seismic data acquisition system. Wireless Geophone Network (WGN) is the acronym that will be used to denote the network of geophones, describing a cable-free seismic data acquisition system. The goal of this paper is three-fold: First, we present a general overview of an onshore seismic survey in oil and gas exploration, underlying the acquisition methodology and requirements. Secondly, the “state-of-the-art” application of WGN in high-density seismic data acquisition and fundamental requirements necessary to set up these networks is outlined. Finally, we considered a subnetwork of wireless geophones in a seismic survey setting, with a focus on the data delivery stage of the survey, and evaluated the network performance based on the geophone-recording period to investigate the optimal number of seismic samples to be transmitted during this period.
Seismic data acquired by a geophone for each sweep can be retrieved at the CCU in either of the following ways: (i) stored internally and retrieved later at the CCU at the end of the survey or (ii) transmitted instantly to the CCU [ 25 ]. The acquisition system is referred to as blind when geophones have no means of transmitting data (i.e., equipment quality control data, noise level monitoring, or seismic data) during the survey to the CCU for monitoring. The equipment quality control (QC) data transmitted could be the status of either the battery level, GPS, storage capacity, sensor (tilt, impedance), etc. Although blind systems offer operational efficiency, the risk of having faulty recordings or data loss is high, thereby compromising the quality of data recorded [ 26 ]. For most seismic exploration companies, it is of significant importance to have a system that provides some QC information and transmits the seismic data acquired during acquisition, often referred to as real-time systems. A real-time system supports the instant transmission of data from equipment in the field to the CCU during acquisition.
Reflected signals from geophones pass through an amplifier and ADC for signal amplification and digitization, respectively, after which it is transmitted to the CCU for further processing. The CCU is also responsible for monitoring geophones as well as timing during the survey. A stream of digitised samples drawn from a single geophone is called a “seismic channel” [ 9 ]. The rate at which data is generated () from one seismic channel depends on the sampling interval () with which the signal of the reflected wave is sampled and the resolution () of the ADC used. Depending on the seismic objective, instrument used, and required bandwidth, the sampling interval typically ranges fromto 8. The lower the sample interval, the higher the frequency range that can be recorded. Above the Nyquist frequency, frequencies are sampled incorrectly. Similarly, the larger the number of bits in the ADC the greater the dynamic range, i.e., the range of signal amplitudes that can be converted without distortion [ 16 ]. Equation ( 3 ) gives the relationship between, and
Seismic sensors used in an onshore seismic exploration can either be geophones or accelerometers. The most commonly used sensors are the geophones [ 21 ], which measure ground motion or velocity when shaken by an energy source and converts it into electrical energy. The moving-coil system design is typically employed in geophones [ 22 ]. It consists of a coil suspended by means of a spring and permanent magnet surrounding the coil. The motion of the ground causes the coil to move through the magnet’s field, thereby producing a voltage that is proportional to the relative ground vibration or the velocity of the coil. Basically, the sensitivity of the geophone is such that a unit ground velocity produces voltage measured in millivolt per inch per second (mV/ips) [ 17 ]. The geophone is a damped resonant system with a damping factor usually in the range of 0.5–0.7 and a resonant frequency usually from as low as 8to greater than 100, which provides adequate coupling for most seismic applications [ 21 ]. Waves travelling through the earth’s surface comprise of three orthogonal components in nature [ 23 ]. A normal geophone measures only the vertical component of the ground’s motion velocity [ 21 ]. A three-component (3-C) geophone, with three spring-mounted coils arranged orthogonally in a single case, measures the three mutually orthogonal components of the particle velocity [ 23 ].
Vibratory: Vibrators (vibroseis) are the most commonly used sources in land seismic exploration [ 18 ] due to some advantages it has over impulsive sources, such as limited frequency band, low-power source, longer energy emission time [ 19 ], less safety concerns, as well as better control of sweep repetition. A vibrator is a vehicle-mounted energy source that converts an electrical signal into high-pressure hydraulic flow to vibrate and control a heavy base plate held in contact with the ground by the weight of the vibrator vehicle and isolated by an air-bag suspension [ 19 ]. A reference sweep signal encoded in the vibrator electronics is sent into the ground through the vibrator plates for a given period of time known as the. Geophones at the surface records the data or seismic reflections for a duration called the 20 ]. The period from the completion of the sweep to the end of recording is the listen time. The vibrator truck lifts the plates and then moves to the next shot point in the survey area. The procedure is repeated throughout the survey area. Vibrators have a typical signal frequency range of 5–511and a sweep length of up to 31 17 ]. One major drawback of vibrators is that they cannot be used in complex terrains, such as mountainous or marshy areas, or in jungles.
In an onshore survey, an energy source (explosives, thumper, vibroseis, sledgehammer, etc.) placed at a shot point generates a low-frequency seismic wave that propagates into the earth’s subsurface through various paths. These seismic waves are reflected and refracted as they experience changes in the geological layering of the earth. A network of seismic sensors or receivers (geophones) deployed in the survey area at the surface detects and records the reflections arriving at the surface during the period of time called “” and converts them into electrical signals, which are then amplified, filtered, digitised, multiplexed, and sent down to the central control unit (CCU) for further processing to obtain a visual image of the earths subsurface structure [ 16 ], after which the source is moved to the next shot point, and the next record is taken. The acquired image is then interpreted by geophysicists or geologists to identify pockets of oil and gas before any test drilling commences. Figure 1 gives an illustration of the seismic survey process.
In sedimentary materials, the elasticity and density strongly depends on the porosity of the layer. S-wave velocities (in the range of 100–500) are much slower in comparison with P-wave velocities, as such are termed secondary waves as it arrives later than the P-wave. P-wave velocities ranges from 200 to 800and 1500–2500for dry and water logged materials, respectively, for sediments such as clay, sand, or gravel. Similar to sound waves, these waves are reflected and refracted as they hit boundaries between different subsurface layers. The magnitude of the difference between boundary densities and seismic velocities defines the intensity of the reflected wave. The seismic wave energy is proportional to the square of its amplitude, and the propagation velocity is the product of its wavelength () and frequency () [ 14 ].
The propagation of elastic waves through the earth subsurface is the basis of seismic exploration. Body waves that are capable of travelling through all distance levels of the earth’s subsurface and Interface waves that found near boundary of layers are the two classes of elastic waves. Most seismic surveys are based on the analysis of body wave signals produced by a seismic source and recorded at the earth’s surface [ 13 ], as illustrated in Figure 1 . Shear or Secondary wave (S-wave) and Compressional or Primary waves (P-wave) are the two types of body waves that exhibit particle motion orthogonal and parallel to the direction of wave propagation, respectively. Using the wave travel velocity and the time required for the wave to return to the surface, the depth of different geological layers could be determined. The wave velocity value carries information on the type of rock or sediment in the subsurface. Depending on the subsurface layer’s density and elasticity, the velocities of these waves are expressed using Equations ( 1 ) and ( 2 ) below:whereis the compressional modulus, µ is the shear modulus, is the density of the subsoil, andandare the velocities of the secondary and primary waves, respectively.
In flat topology sensor nodes have the same functionality in the network and nodes can transmit data directly to the sink in single-hop or multi-hop communication without the need for central coordination. In cluster-based topologies, however, nodes are classified into cluster heads and cluster members. Cluster heads serve as a coordinator and control how cluster members access the shared wireless medium. Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA), a random access scheme based on the exponential back-off algorithm is often employed in flat-based topologies as the fundamental algorithm to access the medium since no coordinator is employed [ 33 ]. In random access schemes, nodes asynchronously contend for channel access whenever they have packets to be sent. The probability of collision increases as the number of nodes or offered traffic increases. Although random access or contention-based approach provides good scalability in WSNs, some of its major drawbacks include collision, idle listening, overhearing and control packet overhead [ 35 ]. On the other hand, most cluster-based networks employ Time Division Multiple Access (TDMA) or schedule-based MAC schemes in accessing the medium [ 33 ]. In such schemes, nodes access the channel in an orderly manner. Resources (time or frequency) are pre-assigned to avoid collisions, as each node has a unique time slot to transmit its data. In the event that a node has no data to transmit, its allocated resource is wasted. To avoid underutilization of resources, schedules can often be changed by re-allocating time slots when needed [ 36 ]. However, this comes with the cost of adding overhead to protocols or schemes. In TDMA, time synchronization is required amongst nodes to avoid overlapping transmissions as well as guard slots to separate users. Although random access schemes are vulnerable to collisions, they are more scalable in terms of static channel allocation as compared to TDMA [ 37 ]. In addition, for a heavy traffic load with a short fixed time slot, the end-to-end delay will be prolonged [ 35 ].
Most exploration companies require that data acquired for a particular shooting phase should be observed for quality control in real-time prior to the next shooting phase. For some systems, seismic data acquired from one sample interval is required to be sent prior to the next sample interval [ 11 ] or within a reasonable amount of time (usually the time lapse between shots) with minimal latency. The network needs to be designed for continuous operation for up to several days, depending on the survey type or size. Information on geophone positioning and elevation is of vital importance for the surveyor to easily keep track of all geophones in the survey in case of failure or damage to ensure the robustness of the system. Most geophones use the Global Positioning System (GPS) for localization and timing. To avoid degradation of the subsurface image quality, accurate positioning with the error below 1 m is necessary after deployment. Seismic reflections from the earth subsurface are simultaneously detected and sampled by geophones in the vicinity of seismic sweep within the survey area. Acquisition needs to be synchronised with the time-reference distributed over the survey area and a maximum acceptable timing skew/jitter of 10–20 μs [ 29 ].
As mentioned earlier, the rate at which data are generated by a geophone depends on the signal sampling interval and the resolution of the ADC used. For a survey with a minimum sample intervalofand a 24 ADC resolution, each geophone in the network will generate data at rate of 48. Data generated by geophones required to be transmitted to the CCU for each record or shot could be quite large. This can be quantified by an example that entails 30,000 seismic channels or geophones, with seismic signals sampled atinterval, 5geophone recording period, and a time lapse between records of 60. For each record, 240of data is generated per geophone, and an overall data ofGbit (900 MB) is generated in the network. To keep up with the data throughput, at least 120bandwidth is required, as well as multiple radio links to share the bandwidth load [ 11 ]. A typical survey with up to 30,000 geophones transmitting data in real-time will require a sum throughput ofGbit/per shot, which will scale down to 720if a sampling interval of 1is used. Support for such throughput is quite uncommon for traditional wireless sensor networks.
Depending on the survey size, a typical land seismic exploration comprises of 10,000 to 30,000 geophones [ 30 ], covering an area in the order of several square kilometres with an average density of up to 2000 devices per km. According to [ 9 ], the amount of traffic or connections in a typical survey is similar to what telecommunication operator handles in a low-populated city. Seismic surveys often employ orthogonal geometry [ 31 ] with source and receiver lines deployed perpendicular to each other according to predefined survey parameters, as shown in Figure 3 . The orthogonal deployment is normally employed to combat the problem of aliasing noise (an effect that causes different signals to become indistinguishable when sampled) with low apparent velocities [ 31 ]. The(also called in-line) consists of geophones placed at regular intervals along the line. The source line consists of shot points marked at regular intervals along the line.are locations on the earth surface where the seismic waves are generated that propagate down the earth’s surface [ 32 ]. WGNs are often static networks as mobility of geophones is not required after deployment. This implies geophones can stay connected wirelessly long enough to support hierarchical networks with autonomous entities [ 9 ].
The growing interest in large-scale seismic data acquisition for better subsurface image quality, flexibility, and automation in surveys [ 27 ], as well as impediments imposed by the cable, is some of the fundamental rationale for moving towards wireless seismic acquisition systems. Such systems often employ geophone units equipped with wireless transceivers to form a wireless geophone network (WGN) deployed over the survey area. WSDA enables a greater trace density and increased productivity, thereby increasing the earth subsurface imaging quality as compared to cable-based systems [ 28 ]. A WGN comprises of a group of wireless geophones that monitor and record ground movements, transmitting their data via radio link to another node in the network, usually a wireless gateway, towards the direction of a sink node or CCU. The gateway acts as an intermediate node and also provides coordination to a group or subnetwork of geophones in the network [ 29 ]. In the following text, we outline some requirements necessary to set up geophone networks for WSDA.
An architecture that defines a serial data transfer path amongst geophones is described in [ 49 ]. Geophones are deployed in series to relay seismic data in a chain-like pattern to the CCU using multiplexing techniques, such as time or frequency division multiplexing. Although the spectrum is effectively utilised, data acquisition time might be high, as well as power consumption of nodes. A wireless architecture based on the IEEE 802.11af standard is proposed in [ 20 ]. In this architecture, the survey area is divided into hexagonal cells based on the frequency reuse concept, such as in cellular communication. The cluster size is determined by the number of available TV white space channels, and each cell consists of a wireless gateway that collects data from geophones within the cell area in a star topology. However, issues of co-channel interference, geophone power consumption as well as regional regulations for accessing TV white space spectrum may be a cause for concern. In the RT3 system of wireless seismic [ 43 ], multiple relay units (gateways) are employed in the network, where a group of geophones (referred to as motes) for a portion of the survey will be associated with and will transmit their data directly to this relay unit. Each relay unit will then relay the acquired data to the CCU wirelessly, either directly or through multi-hop communication with neighbouring gateways. Authors in [ 48 ] proposed a dual-layer network (lower and upper layer network) design for the WSDA system. The lower layer employs a star network topology comprising of eight wireless geophones and a gateway to form a subnetwork, communicating based on 802.11 WLAN technology. In the upper layer network, the gateway in each subnetwork forwards the aggregated data to the CCU via the LTE network. However, a significant amount of gateways will be required with such network architecture.
Conventional orthogonal geometry where receiver lines and source lines are placed perpendicular to each other is often employed in seismic surveys. A receiver line interval and an inter-geophone distance are pre-defined according to the survey parameters before the commencement of data acquisition. A denser receiver arrangement leads to better subsurface image quality. Savvazi et al. in [ 9 ] proposed a hierarchical WGN architecture in which the survey area is broken down into subnetworks, with each subnetwork managed and coordinated by a wireless gateway. Subnetworks are further organised into clusters where geophones connect to the associated cluster heads and transmit their data to. The cluster heads then relay the aggregated data from geophone members to the gateway using short range UWB communication technology and from the gateway to the CCU using long range WiFi technology. However, no performance evaluation was carried out in the proposed architecture. Furthermore, due to the short communication range in UWB technology, a multi-hop network is needed to convey data from geophones, which might be operating near their communication range limit at best. Moreover, most of the data traffic is expected to be relayed via the cluster heads closest to the gateway, raising concern for coordination and energy consumption in such nodes.
There is plenty of research that discuss technologies for wireless seismic data acquisition networks. Authors in [ 12 ] discussed the feasibility of employing WiFi, WiMax, Bluetooth, ZigBee, and Ultra-Wideband (UWB) technologies in WGN. However, considering the WGN requirements in terms of the data rate, network size (in case of Bluetooth), power consumption, real-time data transfer, etc., these technologies might not be the best options. An elaborate explanation of employing UWB technology in WGNs is discussed in [ 9 ]. The authors proposed UWB technology as a reasonable choice at the physical layer considering the advantage of large bandwidth (greater than 500) to provide high-quality delay and a high data rate to support bursty traffic. However, due to a short communication range (5–10for up to 480 Mbit/s, and 30expected for outdoor line-of-sight environment withMbit/s data rate [ 9 ]) confined in UWB, geophones will be operating relatively near their range limit with a multi-hop form of communication to convey seismic data. The design of a wirelessly distributed three-component seismic data acquisition system based on Long Range (LoRa) wireless technology is discussed in [ 44 ]. The system monitors the parameters of seismic waveform, such as sampling rate and signal amplification, in real-time during the survey. The data collected by the system are locally stored on an SD card and then forwarded to the monitoring unit via LoRa technology for further processing. However, LoRa might not be the best candidate for WSDA, especially considering data rate requirements in high-density WGNs. In addition, the authors in [ 45 ] proposed an Internet-of-Things-based Low-Power Wide-Area (LPWA) WGNs with a focus on seismic quality control QC data not a real-time WSDA system. In [ 20 ], the authors proposed a wireless geophone network based on the IEEE 802.11af standard that enables WLAN operation in television white space (TVWS), which can achieve longer communication range, providing more network scalability in WSDA. IEEE 802.11af occupies many various pre-licensed bands TVWS channels, as such its operation might be limited based on potential regional interference and access regulations [ 46 ]. An integrated energy-efficient wireless sensor node for microtremor survey method was proposed by the authors in [ 47 ]. They suggested a data quality monitoring system that solves some technical issues between the recording unit and the control centre through 4G wireless monitoring technologies. Authors in [ 48 ] proposed a wireless exploration system based on a mixture of WLAN and LTE technology. Although LTE can offer considerable data rates, spectrum licensing and deployment costs could be a cause for concern.
In the RT3 system, the acquisition units and gateways are controlled and managed by the CCU. Each recording unit transmits its data to the GRU associated with it using a Time Division Multiple Access (TDMA)-based MAC protocol. The relay unit then forwards the data from one GRU to another in a multihop manner until it reaches the CRS, where the seismic data acquired is processed. The GRU is a full duplex transceiver, which significantly reduces communication latency and allows the RT3 to scale over 250,000 channels with full real-time data transmission. The sustained throughput of the GRU sub-system is approximately 20 Mbit/s with a “burst” rate of up to 55 Mbit/s. The GRU radio network is a fully automated formation sub-system, such that the GRUs self-organise with adjacent GRUs. A GRU is typically mounted on a tripod and, depending on the terrain and user-selected height, the distance separating the GRUs is usually between 100 and 400 43 ].
Owing to its ubiquitous and license-free nature along with modelling basis, the IEEE 802.11 g technology that operates in theISM band is employed. It has a maximum data rate of 54 Mbit/s and utilizes a 22channel bandwidth, making it possible to indulge three non-overlapping channels in theISM band. The 802.11 medium access control (MAC) sublayer architecture defines channel access mechanisms that includes the Distributed Coordination Function (DCF), the optional Point Coordination Function (PCF), and Hybrid Coordination Function (HCF) [ 50 ]. DCF is the fundamental access mechanism in 802.11 in which stations (nodes) contend for access to the medium in a random manner based on the carrier sense multiple access with collision avoidance (CSMA/CA) protocol, where collisions are managed according to binary exponential back-off rules. DCF defines two techniques for packet transmission: the basic access and the request-to-send/clear-to-send (RTS/CTS) mechanisms. Basic access mechanism is the default DCF scheme in which, following a successful reception of packet by the destination station, an acknowledgement is immediately transmitted to the source station. In the RTS/CTS scheme, the source station sends a RTS frame to reserve the channel before transmitting a packet. A CTS frame is sent back by the destination to the source station to acknowledge the receipt of the RTS frame. Subsequently, the normal packet transmission and acknowledgement response occurs. In this work, we limit our investigation to the DCF basic channel access mechanism. The RTS/CTS access mechanism is turned off.
For real-time seismic data acquisition, it is required that data acquired from one sample interval are transmitted prior to the next sample interval [ 11 ] or within a reasonable amount of time (usually the time lapse between shots) with minimal latency. A delay bound of 1if defined as a measure of the QoS in our WGN scenario, within which it is expected that all data from geophones in the network are received at the sink node after the period of time, and the acquired seismic data QC check can be performed for that shot before moving to the next shot point. This mean that we defined a total acquisition time with a lower bound equal to 5and an upper bound of 6(5listen period plus 1). An additional measure of QoS considered is that there should be 0% packet loss in the network.
As the proposed network architecture consists of a number of subnetworks, we focus on investigating the geophone recording period acquisition technique for a single subnetwork only with a defined number of geophones to obtain an initial insight into the network’s performance. Figure 6 depicts the scenario under which the study was carried out. A single-fleet operation with one vibroseis truck, and single shot was assumed. A subnetwork from the network architecture proposed in Figure 4 , consisting of 100 geophones within the communication range of a gateway, was considered. The gateway acts as an access point or relay unit to a sink node, where geophones transmit their seismic data to, as well as provides coordination to the geophones, as defined in the IEEE 802.11 Basic Service Set architecture. The subnetwork consists of four receiver lines placed at intervals of 200. Each receiver line consists of 25 geophones separated by a distance of 25and spread over an area of. The gateway is placed at the centre and connected to the sink node via a high speed cable (100 Gbit ethernet). This is employed to model an ideal connection between the gateway and the sink node in the simulation, as our scenario considers one subnetwork only and the gateway does not relay data to the CCU.
Imagine a survey in whichis defined. This means each geophone will acquire a number of seismic data samples over the period of 1 s, and the corresponding data samples will be placed in a packet and sent over the network prior to the nextuntil the end of the 5recording period. In this example, a total of five packets will be generated per geophone for the duration. However, this also means that 5 times as much channel access will be required as compared to 1 time channel access with one larger packet containing all seismic data samples accumulated over the period. Authors in [ 20 ] have discussed a data collection scheme that can be applied during the listen interval using a TDMA-based protocol, investigating the optimal time slot to be allocated to geophones during listen interval.
However, in this work, we focus on investigating the performance of the proposed acquisition technique for a given number of geophones in a single subnetwork. As most seismic acquisition applications require data to be transmitted after the geophone recording or listen period (), in this work, we propose data be transmitted amid, referred to as “recording period acquisition”. This provides a form of real-time data transfer, thereby significantly increasing the overall seismic survey productivity. A number of seismic data samples () collected by geophones over a given period of time () duringfor a particular sweep is transmitted to the CCU at the end of everyperiod prior to the nextuntil the end of, as depicted in Figure 5 represents the seismic sample accumulation time and defines how frequent packets are sent by geophones in the network. In this work, a typical value ofis used.
To investigate the feasibility of future WSDA applications using the traditional approach and to identify bottlenecks and problems, we proposed a WGN architecture and an acquisition technique. Similar to [ 48 ], the survey area is divided into subnetworks, as shown in Figure 4 . Each subnetwork is managed by a wireless gateway that serves as a central coordinator and data aggregator for a number of geophones that fall within its communication range. Each geophone in the subnetwork transmits its sensed data to the gateway via 802.11 WLAN technology. To limit the effect of adjacent and co-channel interference, neighbouring gateways are expected to use non-overlapping channels (for example, the three non-overlapping channels in theband (channels 1, 6 and 11)), in such a way that no adjacent gateway in a subnetwork uses the same channel. These channels can then be reused in different subnetworks throughout the survey area. The gateway then relays the aggregated data to the CCU via multi-hop communication with neighbouring gateways or via single-hop communication using a longer range communication technology. Data compression techniques could also be applied at the gateways to reduce the high bandwidth requirement or the number of bits transmitted to the CCU. This proposed architecture approach could be quite promising in terms of scalability and better network planning and management. The gateways should have a dual interface to be able to support both short and long range wireless technologies.
In comparison with the measured throughput from simulation, it can be seen that the throughput closely follows the offered load up to 15payload size. Beyond 15, the throughput starts to drop significantly. This may be explained by the fact that packets get fragmented at the network layer once the payload exceeds the MTU size, resulting in additional smaller packets (fragments). Each of the fragmented packets have their own header, which results in an increase in overhead and consequently a decrease in throughput. Furthermore, each of the fragmented packets contend to access the wireless channel, which consequently increases network congestion at the link layer. After several retransmission attempts by nodes, some of the packets are dropped, owing to the fact that the retransmission attempt limit is reached, as can be seen in Figure 8 . Throughput at 30payload, which represents the entire seismic data recorded for the duration, tends to be significantly higher. This is because each geophone generates and transmits 30UDP packet only once over the network, which is not the case for lower packet sizes, but gets fragmented into smaller packets at the network layer.
In this section, we evaluate some of the WGN key performance metrics from the simulation for different payload sizes Figure 7 shows the plots for the offered load to the network as compared to the throughput measured in the simulation. The offered load is computed analytically using Equation ( 8 ). It can be observed from the figure that the offered load increases with increase in payload size. This is because the sample accumulation duration () increases with a higher payload size, consequently reducing the time it takes to receive all data packets at the sink node () over the 5geophone recording period duration (). The lower the, the higher the offered load. Asfor both packet size 27and 30are nearly the same, the offered load for both packets sizes is almost the same, as can be seen from the figure. Furthermore, with larger packet size, this would mean more available data in the network and also higher throughput, as long as the capacity is not exceeded.
An initial objective of this study was to determine an optimal value of
n
(a function ofp
) that could be sent during the geophone recording period that ensures the best network performance in accordance with the defined QoS in our WGN setting. The results of this study indicate that the throughput can be maximised by sending the largest possible value of seismic data sample sizen
or payloadp
over the network. This, however, comes with a price of fragmentation at the network layer due to the limit on the WLAN maximum transmission unit (MTU) size (1500 B). The larger the value ofn
, the higher the load on the channel, thereby increasing the chances of packets being dropped, leading to decrease in the throughput. In addition, this will also lead to a greater number of retransmissions, thereby degrading the overall network performance in terms of delay and packet loss. To determine the optimal value ofn
, a comparison between throughput, delay, and packets dropped from the results obtained was carried out in relation to our predefined QoS parameters in this study (1 s upper bound delay and 0% packet loss). Based on this, we defined n = 4000 (which corresponds to data payload size of 12 k B ) as optimal because it has negligible packets dropped, higher throughput compared to smaller values ofn
, and also delay falls within our defined bounds. However, it is important to note that this study has a number of limitations. Firstly, a WLAN MTU size of 1500 was considered, which implies that the optimal value ofn
might vary for different MTU sizes. Secondly, we assumed a free space path loss propagation model. Switching to a more realistic channel model such as the two-ray propagation model will deem more appropriate as geophones often have antenna heights of less than 1 m . Furthermore, the study was carried out with only 100 geophone nodes and one gateway for a single subnetwork under ideal channel conditions. Reducing the number of geophones in a subnetwork will, however, increase the optimal value ofn
(orp
) as a result of the reduced contention at the MAC layer. Nevertheless, the objective is to utilize the maximum number of geophones within a subnetwork, which gives the expected network performance, thereby reducing the overall number of gateways used in the survey. Future work will take the above limitations into account.Want more information on Wireless Seismic Geophones? Click the link below to contact us.