Detecting Intrusion in Large Farm Lands and Plantations in Nigeria Using Virtual Fences

.
Keywords: Intrusion detection, Virtual fences, Motion sensors, IoT, Micro controllers, Monitoring

Security of farmlands in Nigeria is still a major challenge, as most farm owners are constantly faced with issues of curbing theft both from humans and animals alike. Common solutions might include building brick fences around the farmland, installing electric fences or plant deterrent plants with spikey branches or those which give out displeasing scents. Though relatively effective, these methods can be less than ideal and are sometimes prohibitively expensive to put in place.

Even when installed these conventional approaches to fencing have some inherent shortcomings prominent of which is being easily crossed either with the use of ladders, ropes or other means. These types of fences also provide shades for lurking intruders to hide while monitoring activities within the farmland or plantation prior to perpetrating their plans. Also being made of brick and mortar, holes can easily be bored into these fences, through which intruders can easily crawl in.

In developing countries and especially rural areas, these fences are often made with sticks and ropes and are usually the only security measure put in place by the plantation owner. They are easily crossed and the intruders can cart away with as much crops as they can carry without the knowledge of the owners. Infrastructure and site surveillance is therefore imperative as the security and safety of crops is an ever growing source of concern.

To overcome this challenge, we propose the use of a virtual fence. A virtual fence has been defined as a bounded enclosure around a farmland without the use of any physical barrier (Umstatter, 2010) and it is an application of Internet of Things (IoT) in smart farming (Beecham Research, 2014). IoT has been defined by the Internet Architecture Board, to be a trend in which numerous embedded devices communicate over Internet Protocol (RFC7452, 2015). It is a network of objects that have been embedded with components that enable them retrieve and exchange data over an IP network with little or no human intervention (ITU, 2015). It comprises of four models viz: device-to-device, device-to-cloud, and device-to-gateway and backend data sharing and has numerous applications in monitoring, control, automation, health care, wearable and smart systems.
This paper focuses on the application of IoT in monitoring (Gupta, 2015) and applies the device-to-cloud model in creating the virtual fences. These fences rely heavily on the use of active sensors (Felemban, 2013) for its operation and can be installed in place of or alongside physical fences to form a perimeter around the farmlands.

The rest of this paper is structured as follows, section two presents a survey of related works, section three discussed our proposed approach, section four discusses our experimental setup, while discussion on results obtained is in section five, conclusion and future works are discussed in section six.

A general review of the use of IoT for monitoring home appliances as well as active devices in computer networks was done in (Gupta, 2015). The author also did a comparison between two monitoring agents – Big Brother and Zenoss and it was concluded that Zenoss was the better of the two due to some added features which the former lacked.

In the work carried out in (Felemban, 2013), a perimeter wall was built using multiple pairs of infrared sensors around an area of interest. Each pair consists of a transmitter and a receiver connected wirelessly to a centrally monitored control station. Once the infrared link between the transmitter and receiver is broken by an intended intruder, an alert signal is triggered at the control station notifying the presence of an intruder.

The authors in (Chan, et al., 2014) a performance analysis of virtual fences using Wireless Sensor Networks (WSN) was presented. This work used the Doppler effects (Evans & McDicken, 2000) to detect motion along the virtual fence and Radio Frequency (RF) signals to transmit information to the central control. Sensors were placed at different positions both indoor and outdoor with measurements taking to determine the maximum sensing range at these positions. Various criteria were considered some of which included the Azimuth angle, sensor height and sensitivity levels. It was concluded that Doppler effects based motion sensors were best suited for indoor use as against outdoor use where waving trees could be confused for intruders by the system.

In the work done in (Vermani, 2013), the authors presented the use of virtual fences for managing grazing animals. A pair of transmitter – receiver Radio Frequency modules is employed. The transmitter is placed around the collars of grazing animals while the receiver is installed at the monitoring center. The transmitter continues to send signal to the receiver as long as the animal remains within range. Once the animal crosses the preset range, transmission stops and an alarm is consequently triggered to notify the owners. Though seemingly effective, the long term effect of prolonged exposure to RF signals on the animals is not considered. The proposed approach is also a reactive process rather than a proactive one and does not actively prevent the animals from wondering outside the preset grazing area.

A comparative survey of four fencing techniques was done by the authors in (Goyal et al., 2012). These techniques include conventional brick fences, electric fences, virtual fences using GPS and virtual fences using RF modules. The authors concluded that the RF and GPS based virtual fences were both flexible, easier to setup, and cheaper in the long run. It was however noted that the conventional fence still had its place especially when the area to be fenced off is close to hazardous environments such as highways. The author however did not consider the health implication and the range of RF modules; the actual cost of brick fences versus GPS based solutions on a large scale and the accessibility of GPS and RF modules versus conventional brick and mortar conventional fences especially in rural communities

In the work of (Butler, et al., 2006) a combination of robotics, animal behavior and network were combined to create a controlled virtual fence. Their work focused on creating a fence-less way of guiding herd of cows to feeding areas along a pre-mapped route using a dynamic virtual fence that can be moved as desired. In order to ensure the grazing animals comply with these mapped routes, scary sounds such as dog backs or lion roars are used as guide. The main advantage of this approach is that the virtual fence can be moved easily across and around grazing areas. Their results also show that the use of sounds as external stimuli was highly effective when used on cows. However most of the experiments were done using simulation tools and in some cases on humans; this introducing errors as a result of predisposed biases.

Article III.The Proposed Approach

In developing our system, two sensor modules were placed along the perimeter of a lawn. A lawn was used because in the immediate a large farmland was outside our direct reach and the lawn approach could serve the same purpose. Our experimental setup therefore involved simulating the installation along the perimeter of a farmland by deploying the sensor modules along the boundary of an open lawn having a “Do not walk on the lawn” sign post. This is as illustrated in Figure 1. In Figure 1, the sensor modules are placed along the edge of the lawn where the walkway splits

Figure 1: Experimental deployment of a virtual fence along the perimeters of a lawn

We chose this area because people are fond of walking across the lawn, rather than on the walkway around it. This we considered as being analogous to intruders coming in through wrong paths.

Figure 2: Micro Controller based Sensor Module

A sensor module prototype was constructed using a combination of an Arduino Micro Controller, WiFi Shield (ESP8266-12) to provide Internet connectivity to the Micro controller, a PIR motion sensor and a RTC module. These combined hardware are simply referred to as the sensor module (see Figure 2). The Arduino Controller is based on the ATMega328P with 54 digital I/O pins, 256Kb of Flash Ram and 8K of SRAM. The WiFi shield was then placed on the Arduino and provided a means of wirelessly accessing the Internet. The PIR sensor module used is a HC-SR501 model, which was powered from the Arduino micro controller and connected to PIN 5. The RTC used was the DS1307 model a simple and efficient I2C based Real Time Clock and was used to keep log of time.

In order to communicate with the WiFi shield, the WiFiEsp Library (Bportaluri, 2015) was used and set to a serial baud rate of 112500. This provided access to the Internet, making our sensor a web client that updates a database. Though the database was setup on Microsoft Azure– a Cloud PAAS platform (Microsoft Azure, 2015), the Sensor was not directly connected to it. Rather the sensor updates a local database, which in turns syncs with the remote database. This was necessary as the Arduino does not support SSL connection, which almost all Cloud database require. Programming for the Sensor was done in Arduino sketch which is a variant of C/C++.

Each sensor module was mounted inconspicuously at the edge of the area of interest. The sensor module was highly sensitive with a detection range of about 3 meters. Once a person moves across its field of view, the sensor detects the passage and sends a “HIGH” signal to the micro controller. This in turn updates the remote database and the corresponding timestamp recorded

Our experiment show that people indeed walk across the lawn and these passages were detected and recorded by the sensor module as depicted in Figure 3. It takes less than a minute to walk across the entire area and as shown in Figure 3, record 3 and 4, 7 and 8, 9 and 10 indicate a complete walk across. Other results indicate that such individuals crossed at either of the sensors but possibly exited along the other perimeter of the lawn.

These results though not the exact test location, show that the use of sensors similar to ours can be used to detect intrusion and logged accordingly for future analysis. Relevant information such as time of the day, location and possibly distance can be obtained from the data logs. Over a period of time and with a large number of sensors, the data gathered can be extremely large. Information Retrieval tools such as those reviewed in (Accomazzi, et al., 1995 and Onwuchekwa & Jegede, 2011) can be used to polled data and extract relevant information which can then be used to make intelligent decisions; such as when and where to intensify security measures in a bid to cub theft or other related malicious activities being experienced around the plantation.

Figure 3: Snapshot of data log exported to Microsoft Excel

The use of Virtual fences to monitor the perimeters of large farmlands and plantations can be of tremendous advantages. The most obvious is in terms of cost savings when compared to building high brick fences and employing security personals to patrols the entire perimeter. With virtual fences, relatively cheap modules can be installed and the entire perimeter monitored remotely

In this work, we have simulated a virtual fence for plantation and though our obtained results show a lot of promise, there is room for improvement. For instance, as a way of improving the efficiency of the system, a motion sensor with greater range could be used. A sonar based sensor for instance could be used in place of the PIR sensor as the former also give information about the exact distance the intruder is from the sensor. More sophisticated sensor units such as those that measure heat signatures can be used to determine if the intruder is human or animal, unlike ours that simply detected intrusion but could not specify the type of intruder. Multiple sensor modules can also be combined to allow coverage of long perimeters.

A major weakness of the IR based sensor is that it requires direct line of sight. This implies that in the presence of obstacles it would not suffice. To deal with this, RF based motion sensors could be used in place of the IR sensors as these have the advantage of being able to propagate through obstacles such as walls and tree trunks.

A GSM/GPRS module could also be incorporated into the sensor, to provide cheaper and wider access to the Internet via the use of mobile Telco SIM cards for data logging as against the WiFi based system used in our work.

List of References

Accomazzi, A., Murtagh, F., Rasmussen, B. (1995) Information Retrieval Tools and Techniques. Available at https://www.eso.org/sci/libraries/lisa/lisa2/papers/accomazzi-murtagh-rasmussen/fionn-murtagh.pdf (Accessed 7/07/2016).

Beecham Research (2014) Towards Smart Farming: Agriculture Embracing the IoT Vision. Available at https://www.beechamresearch.com/files/BRL%20Smart%20Farming%20Executive%20Summary.pdf (visited 26/05/2016).

Bportaluri (2015) Arduino WiFi library for ESP8266 modules, [Computer Software] Available at https://github.com/bportaluri/WiFiEsp.git/ (visited 24/05/2016).

Butler, Z., Corke, P., Peterson, R. and Rus, D. (2006) From Robots to Animals: Virtual Fences for Controlling Cattle. International Journal of Robotics Research, Vol. 25, No. 5-6, pp 485-508.

Chan, H., Rahman, T. and Arsad, A. (2014) Performance Study of Virtual Fence Unit Using Wireless Sensor Network. 8th Intl. Conference on Sensor Technology, Liverpool, UK, pp 534–537

Evans, D. and McDicken, W. (2000) Doppler Ultrasound, 2nd ed., New York: John Wiley and Sons.

Felemban, E. (2013) Advanced Border Intrusion Detection and Surveillance Using Wireless Sensor Network Technology, Intl. Journal. Communications, Network and System Sciences, Vol. 6, pp 251-259

Goyal, V., Mudgil, A. and Dhawan, D. (2012) Design and Implementation of Virtual Fencing using RF modules, Intl. Journal of Engineering Research and Technology.

Gupta, U. (2015) Monitoring in IOT Enabled Devices. arXiv preprint arXiv:1507.03780.

ITU (2015) Internet of Things Global Standards Initiative. Available at http://www.itu.int/en/ITU-T/gsi/iot/Pages/default.aspx. (visited 26/05/2016).

Microsoft Azure (2015) What is Azure? [Computer Software]. Available at https://msdn.microsoft.com/en-us/library/azure/dd163896.aspx (visited 29/05/2016).

Onwuchekwa, E. and Jegede, O. (2011) Information Retrieval Methods in Libraries and Information Centers. An International Multidisciplinary Journal, Ethiopia. Vol. 5 (6) No. 23, pp. 108-120.

RFC7452 (2015) Architectural Considerations in Smart Object Networking. Available at https://tools.ietf.org/html/rfc7452 (visited 3/05/2016).