Energy balanced-based data transmission with mobile sensors

Abstract

The mobility of mobile sensors in wireless sensor networks (WSN) is important and widely discussed issue with variety of researches. In many studies, using mobile sensors to help data transmission has been proven to prolong the life cycle of the entire network. But most of the current researches assume the energy of the sensors is unlimited. However, the energy are limited in the real environment, we need to consider the energy limit. So based on the energy consumption, we proposed an EBDTM to give thought to the problem in this paper. After the use of mathematical formulas to set out our issue, constructing a generation path which can as much as possible to satisfy with our energy setting,experimental simulation shows that the proposed mobile mechanism is able to make the network environment has a little longer network life. In this paper, we used a linear regression method to obtain a surrogate path for the time being. In the later extension we will consider using a quadratic linear regression to see if a second regression can perform better than a regression.

Keywords

Energy balance mobile sensor wireless sensor networks network lifetime

1. Introduction

Wireless Sensor and Mobile Networks (WSMN) is a hybrid system, which integrates Wireless Sensor Networks (WSN) and Mobile Sensor Networks (MSN). Through this interactive network, wireless mobile sensor networks can be applied in a given network environment for a basic event monitoring. When there is an event that requires a great deal of attention being paid, the sensor at the periphery of the event can sense the occurrence of the event and issue a request to the wireless mobile sensor network system. The system sends a mobile sensor to the location to collect more data. And the collected data needs to be sent back to Base Station to an operator or decision maker for effective analysis or follow-up. Based on Base Station, the information that is sensed by many sensors is received, the decision maker at Base Station will make the appropriate decision, moving and controlling these mobile sensors to execute by tele-operating.

Figure 1.

Event scenarios.

Figure 2.

Mobile sensor scenarios.

As shown in the example in Fig. 1, the data is transferred from the event point area to Base Station, which will go through the static sensors of 1, 2, 3, 4, 5, but the static sensors closer to the Base Station need to help a lot of other sensors transmit data resulting in the speed of consumption power faster than others. In the Fig. 2, the electricity of the No.4 static sensor is less than the rest of the static sensors. As a result, it is necessary to place a mobile sensor A near to the static sensor, with the help of mobile sensors to extend the network life cycle.

Therefore, in this thesis, we discuss how to move the sensors to a suitable location and to ensure that these mobile sensors are connected. As we mentioned before, Base Station not only can communicate with the static sensors, but also to collect the data received by the mobile sensor. However, the transmission of many data will result in the rapid electricity consumption of static sensor through the static sensor, this is a serious problem. In order to slow down the issue, most of the data transmission should be done by mobile sensors, as the mobile sensor has more power compared with the static sensor. And whose communication radius is also much longer than the static sensor. Integrated above, to be able to more efficient use of power resources, as well as to prolong the overall network lifetime, we should let the mobile sensor to help carry the large amount of data, the lighter data is supposed to be given the static sensor to do transmission.

Therefore, whenever there is a large amount of transmission event point occurring, we will need the mobile sensors which must be able to serve the entire network to assist the generation of data.

In addition to the mobile sensors that were used for establishment of network connectivity having been added, we also need to consider the idle mobile sensor that has not yet been assigned to the task. Of course, we can let them stay in the current position, but once the mobile sensors, participating in the establishment of the communication path, move to a new location, some other mobile sensors not assigned to the task may not be connected through the MSN system Base Station. To solve the problem, we can use the WSN and those mobile sensors to build connection; utilizing WSN to send control commands and new tasks for these mobile sensors, this method is feasible and makes the resource usage of the network more efficient.

Then one of the obviously goals is as far as possible to reduce moving distance by the mobile sensors. We also can try to minimize the number of hops to achieve the reduction of the number of mobile sensors so that the mobile sensors which are not assigned the task have the opportunity to implement other tasks.

Maintaining the network connectivity of the Base station and Event happening Point was viewed as a problem of Mixed-Integer Quadratic ally Constrained Quadratic Program Restrictions (MIQCQP) [2], it is a problem for NP-Hard [1] what there is not the most efficient way to solve, so it is necessary to use the heuristic algorithm [14]. Since the final path based on the previous limitations, we not only devote to minimizing the data transmission hops in this thesis, but also consider the moving distance and cost of electricity which is a pairing problem. To decrease the moving distance and spending, we used the Hungarian algorithm [3].

After the computing of the simulation experiments, we found that the proposed algorithm can effectively reduce the moving distance of the mobile sensor, and there are a little fewer the number of hops between Base Station and event point, so the EBDTM can better prolong the network lifetime.

The other structures of this paper is organized as follows: in the third chapter, put forward the network environment, the assumption and description of the problem; in the fourth chapter, apply some previously proposed background knowledge to this thesis, and then do a detailed explanation of the method forwarded in this paper, the fifth chapter will simulates and compares with the ways of other papers, the sixth chapter summarizes the results of the study and the future improvement goals.

Figure 3.

Communication range of mobile sensor.

Figure 4.

Original sensors operation mode.

2. Related work

In the previous related researches [4], the contend of these studies are on using a mobile sensor to assist data generation [8] to extend the network life cycle, using the alternative mobile sensors and assigning these selected sensors to help the transmission of data by Base Station. As shown in Fig. 3, the research way of the paper is do online directly by using the two point of event point and Base Station firstly, secondly using the communication radius of the mobile sensor shown in Fig. 4 figure out the numbers of mobile sensors to do data generation on online between two point that need to be placed. Finally, Base Station chooses by utilizing Hungarian Algorithm all the mobile sensors of the network environment. The selected mobile sensor will be assigned to a specific location. But when calculate the number of the placed a few mobile sensors in the use of communication radius, if failing to set out, for example, appeared to place 3.8, because there is no way to place the non-integer, mobile sensors must therefore be placed four, so in this case the mobile sensors would need to take more road to the Base Station and the event point online. As shown in Fig. 4, the communications radius of mobile sensors is loose which could lead to the mobile sensors move more distance. When the sensor moves to the specified location, it would have not enough power to continue to do transmission. In Fig. 5, we make the circle in communication range of the moving sensor to be cut at the point where the sensor is moved to achieve a reduction in the moving distance of the moving sensor. In addition, the study does not take into account the power consumption of the mobile sensor, so as we choose the mobile sensor, the electricity will be considered.

Figure 5.

Adjusted ideal position.

Figure 6.

The network environment scene of WMSN.

3. Network environment and the problems description

3.1 Network environment

As shown in Fig. 6, network environment scene of WMSN consists of Static Sensors, Mobile Sensors that both have positioning function, Base Station and event point. Randomly generating event point is divided into a large amount of data and a small amount of data. However, in the previous section we have discussed that in order to make better use of power resources and extend the network life cycle, we hope to transfer a lot of data (such as, photos or movies) through the mobile sensor to establish a connected transmission path which is passed from the event point to Base Station. As for light data transmission (such as, temperature, pressure), we use the static sensor directly for transmission from the event point back to the Base Station.

And if there is a large amount of data which need to be transferred, we firstly let the mobile sensors, not involved in the generation tasks, in the original position, temporarily without any movement, reduce the mobile sensor in the network scene to move so that decrease the electrical consumption of mobile sensors. Next we are going to consider how to establish a transmission path using the smallest number of hops, that is, hope that the data can from the event location back to Base Station with fewer mobile sensors. In this way not only can reduce mass power consumption, also can solve the delay of data transmission.

Figure 7.

The initial state of mobile sensors.

Figure 8.

The new location of mobile sensors.

To transform our problems into the representation approach with system, we make a problem reducing the total movement distance of mobile sensors at first. So we assume that the Base Station know itself current location and the static sensors are also able to use Data flooding mechanism (Data flooding) [13] to let Base Station know each Static Sensor’s current location. When there is a need to establish a transmission path, first of all the static sensors will come back the information of the location of the event point. Then the Base Station would broadcast the location of the event point toward the event point in the form of an area. If mobile sensors receive the radio, they will provide the location information with the Base Station. Next the Base Station will also transmit the calculated way of movement to the mobile sensors. Finally mobile sensors will move the location of arrangement after receiving the message. Figure 7 for the initial state, mobile sensors and static sensors are arranged in a random way, whereas, in Fig. 8, mobile sensors are moved a new location after Base Station allocation.

3.2 Problems description

the article assumed that the P static sensors were built in random, K mobile sensors were built in $\{(x_{1},y_{1}),(x_{2},y_{2}),\ldots,(x_{k},y_{k})\}$ , and these mobile sensors form a multiple hop network environment [11] with event point whose location lie in $(e_{x},e_{y})$ , here we directly let static sensors do the transmission of lighter weight event point back to Base Station, so for the movement of mobile sensors us needn’t to give thought to the situation where using mobile sensors to help transmit the light data.

Figure 9.

The scope of communication if two mobile sensors are able to communicate.

Base Station will broadcast mobile instruction to mobile sensors after the calculation, mobile sensors will determine whether them needs to move to a new location or remain on standby in the current position according to the result. If the action of moving occur, the new position of the mobile sensors are in $\{({x_{1}^{\prime},y_{1}^{\prime}}),({x_{2}^{\prime},y_{2}^{\prime}}),\ldots,(% {x_{q}^{\prime},y_{q}^{\prime}})\}$ . We hope to prolong the life cycle of static sensors by utilizing the mobile sensor to do the generation of data. So it can be divided into two aspects. On the one hand, reduce the average distance of mobile sensors, on the other hand, maximize the life time of static sensors which is survive with the least energy in the network environment scene [10].

To achieve the target of reducing the average moving distance of mobile sensor, we need to use the new position and the original position of mobile model [7]. As shown in Eq. (1), the original location of mobile sensor coordinates $({x_{k},y_{k}})$ and the position coordinates $(x_{k}^{\prime},y_{k}^{\prime})$ after movement, we can obtain the line distance between the two points by placing the two coordinates into the Eq. (1). As we want to reduce the average moving distance, we can utilize the Eq. (2) to represent the whole average distance.

$\displaystyle{D}({x_{k}^{\prime},x_{k}})=\sqrt{({x_{k}^{\prime}-x_{k}})^{2}+({% y_{k}^{\prime}-y_{k}})^{2}}$ (1) $\displaystyle\textit{Minimize}\left({\frac{\sum\nolimits_{i=1}^{k}{{D}({x_{k}^% {\prime},x_{k}})}}{k}}\right)$ (2)

In the Eqs (3) and (4) setting the current position of the mobile sensor, we write the final position to ensure that finally the mobile sensors will move to the position given by the Base Station.

$\displaystyle x^{\prime}_{k+1}=x_{k+1}$ (3) $\displaystyle Y_{k+1}^{\prime}=Y_{k+1}$ (4)

In addition to ensure that the adjacent mobile sensors can communication and let the data from the event point back to the Base Station after final moving, we must limit the communication range of two adjacent assigned move sensor until it is smaller than their biggest communication range (Fig. 9).

While in the goal that maximize the survival time of the whole network environment, we also use mathematical functions to express. For example the Eq. (5).

$\displaystyle\textit{Maximize}\left(\min\left(\frac{\varepsilon_{i}}{cl_{i}+ch% _{i}}\right)\right)$ (5) $\displaystyle cl_{i}:\text{average energy consumption of static sensor $i$ in % light traffic}$ (6) $\displaystyle ch_{i}:\text{average energy consumption of static sensor $i$ in % heavy traffic}$ (7) $\displaystyle\varepsilon_{i}:\text{the remain energy of static sensor $i$}$ (8)

We would like to use a few mobile sensors to help transmit data, so when choose the numbers of sensors to be used, we make use of the sensor’s communication range to calculate, the given numbers of mobile sensors are $H=\left\lceil{\frac{D(\textit{base}_{i}E)}{r_{c}}}\right\rceil$ by which we can know that how much mobile sensors would be needed at least to make data from the Base Station back to the event point, $r_{c}$ is for communication of the shortest distance between two sensors, $D(\textit{base}_{i}E)$ is for the distance between base station and event point.

Figure 10.

The example of the liner regression.

Figure 11.

The line of linear regression environment.

4. EBDTM algorithm

In this section, we proposed the EBDTM algorithm aiming at WSMN. The first part of the preliminary knowledge is a brief introduction on a linear regression [5] and the special properties of the Hungarian algorithm [15]. Next the algorithm will be divided into four stages to introduce: (1) the initialization of the network scene, (2) the monitoring phase of the network scene, (3) the determined stage of position of mobile sensor, (4) the selection stage of the candidate sensor. Finally, Base Station broadcasts the mobile message and moves the sensor to the assigned location to achieve the goal of extending the network lifetime. The details of the various phases will be described below.

4.1 Preliminary

In front of this part we need to explore how to use linear regression [6]. linear regression have a special characteristic that the line segment drawn by linear regression can determine the total distance of all linear regression points to the linear regression line is the shortest. The following Fig. 10 for example, using 5 point to do linear regression L, the line segment drawn out after using the linear regression can always ensure the square of the whole distance of five points to L be the shortest. As shown in Fig. 11, in doing linear regression, we put the candidate sensor and Base Station and event point into regression, we can find a regression line L, and so the sum of the square of the distance between the mobile sensor, event point and Base Station to L is always the shortest. We are going to get this line in the initial advance using a linear regression method, later, we will use two phrase linear regression [12] in the extension to see if the two regression [11] can have better a performance than the linear regression.

Figure 12.

The example of Hungarian algorithm.

Figure 13.

Static sensors in a connected network environment.

The Hungarian algorithm is utilized to deal with $m$ objects to match the $n$ objects ( $m\geqslant n$ ) and minimizes total cost of moving. The following Fig. 12 for example, there are three (A, B, C) mobile sensors and two candidate points (P1, P2). After calculation of Hungary algorithm, we found that the mobile sensor C to the candidate point P1 and mobile sensor B to the candidate point P2 will have the minimum cost.

4.2 Initialization phase

In the whole network environment, as shown in Fig. 13, mobile sensors and static sensors is randomly distributed in the monitoring area. Base Station collects and broadcast location information of static sensors from monitoring area. The network connectivity state is established by these static sensors. So the message of location can be transferred back to Base Station. Next enter the analog phase of the network scene.

4.3 Network scene monitoring phase

The analog phase is divided into two cases. When static sensors sense surrounding events, if the amount of information transmission is light (such as, temperature, pressure), then transfer the data back to the Base Station through the static sensors, as shown in Fig. 14. If it is a heavy amount (such as, photos, videos), after the static sensors send a request of Proxy instructions to Base Station, enter the determined stage where mobile sensors how to place.

Figure 14.

Direct transmission to Base Station by static sensors.

Figure 15.

Base Station broadcast for event point with $\alpha$ angle.

4.4 Mobile sensor selection phase

The first do the calculation that there are at least H mobile sensor which can transfer data between Base Station and event point. Secondly the starting broadcast angle of the sector [9] is determined by the direction of Base Station towards the event point. As shown in Fig. 15, collect location information of mobile sensors in the broadcast area, if the number of mobile sensors is less than H, then increase the broadcasting sector angle: if the broadcast angle has already reached 360 degrees, the area does not have mobile sensor enough to do the generation, Base Station will be informed by static sensors. if the number is more than or equal to H, choose H mobile sensor to draw a linear regression between Base Station and event point, then acquire the regression equation of L. Such as Fig. 16, Base Station to event point requires at least 5 mobile sensors, so select 5 mobile sensors to do a linear regression, a linear regression line segment is obtained.

Figure 16.

Linear regression.

Figure 17.

Base Station in the radius tangent.

Figure 18.

Choose the H candidate points.

As shown in Fig. 17, calculate the first candidate point tangent to L using the circle equation with Base Station as the center and $r_{c}$ as the radius. And then compute the second candidate point tangent to the L using the first candidate point as the center. such as Fig. 18. Repeat the operation until get the number of tangent to the L is H. If the distance from the Hth tangent point to event point is less than or equal to $r_{c}$ , $f$ the H-tangent point and the event point distance are less than or equal to $r$ , the mobile sensor candidate point on the linear regression line L can create a path that causes the data to be passed back from the event point to the Base Station. Then enter the last stage for candidate nodes selection. Assuming that the distance from the last candidate point to event point exceeds $r_{c}$ , it is necessary to translate the linear regression line segments towards the direction Base Station and event point online.

4.5 The selection of candidate sensors

Hungarian algorithm is used to match the H mobile sensors to the H candidate points, making the mobile sensors have the smallest moving cost. And the results calculated are broadcast for the mobile sensors selected by Base Station. When the mobile sensors receive these messages, they will be automatically moved to the candidate point, as shown in Fig. 19.

Figure 19.

The final movement result of mobile sensor.

Figure 20.

The average number of hops in a uniform distribution.

Figure 21.

Average moving distance of a mobile sensor under uniform distribution.

5. Simulation experiment

In this section we compared our effectiveness with the methods we have referred to in the paper. In a range of a 2000 m * 2000 m, place the Base Station at (0, 0) and event point most likely to come from (1000, 1000) or ( $-$ 1000, $-$ 1000). In the part of the mobile sensor we set the transmission radius to 300 m, and the number of mobile sensors used is 25, every 5 as a delivery.

As shown in Fig. 20, when the average number of hops in a uniform distribution in the experimental environment, with the increase of the number of mobile sensors, Hop count of MDRM method increased significantly, our method remains unchanged.

As shown in Fig. 21, when average moving distance of a mobile sensor under uniform distribution in the experimental environment, with the increase of the number of mobile sensors, moving distance of MDRM method lower significantly, our method slow down.

As shown in Fig. 22, when the average moving distance of the mobile sensor under no uniform distribution in the experimental environment, with the increase of the number of mobile sensors, our method has a shorter average moving distance.

The experimental results are shown in Figs 20 and 21, when most of the mobile sensors are evenly distributed in the experimental environment. The result is not much difference between the two methods, but we can still have the minimum number of hops, at the same time also can reduce mobile cost. While in Fig. 22, the experimental environment is uneven; our method still has a shorter average moving distance.

Figure 22.

The average moving distance of the mobile sensor under no uniform distribution.

6. Conclusion

Based on the model of wireless sensor networks, this paper proposed a moving mechanism of EBDTM that discussed how to move mobile sensors to make the total moving distance least, also can transfer data from the event point to Base Station. Among EBDTM we use the method of linear regression to find the candidate points in the linear regression line so that the consumption of energy and the moving distance are the least when mobile sensors need to move to the candidate point. Then the mobile sensors can save much energy to do data transmission. The simulation results show that using the mobile system of EBDTM can make the moving distance less and let the lifetime of whole network longer.

7. The future research

In the paper, future picture of the research work and development are as follows: Initialization layout of mobile sensor can be intelligent prediction, the initial position of mobile sensors as far as possible is reasonable. The design of mobile sensors can use solar battery; increase the service life of mobile sensors.

Footnotes

Acknowledgments

The research is supported by the Natural Science Foundation of Department of Education of Anhui Province (No. KJ2017A325)

References

Bevern

R.V.

, Fixed-parameter linear-time algorithms for NP-hard graph and hypergraph problems arising in Industrial Applications, 2014, 15–43.

Lee

and Leyffer

, Mixed-integer nonlinear programming, Springer New York 9(4) (2012), 329–349.

Kuhn

H.W.

, The Hungarian method for the assignment problem, Naval Research Logistics 52(1) (2005), 7–21.

Wichmann

Korkmaz

and Tosun

A.S.

, Minimum hop and/or minimum distance robot movement with connectivity constraints in WSRNs, Global Communications Conference, IEEE (2014), 523–529.

Zhu

Nallamothu

and Ryu

, Impact of linear regression on time synchronization accuracy and energy consumption for wireless sensor networks, Military Communications Conference, 2008 Milcom, IEEE (2008), 1–7.

Austin

P.C.

and Steyerberg

E.W.

, The number of subjects per variable required in linear regression analyses, Journal of Clinical Epidemiology 68(6) (2015), 627.

Nabi

Geilen

and Basten

, MoBAN: A configurable mobility model for wireless body area networks, International ICST Conference on Simulation TOOLS and Techniques ICST (Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering) (2011), 168–177.

Neill

K.O.

, Applying a volume dipole distribution model to next-generation sensor data for multi-object data inversion and discrimination, Proceedings of SPIE – The International Society for Optical Engineering 7664(1) (2010).

Sumi

Komiya

Yamaguchi

and Cho

, Design of base station antenna using hybrid tilting, Ieice Technical Report Antennas & Propagation 109(218) (2009), 67–72.

10.

Yin

et al., Visual analysis and simulation of dam-break flood spatiotemporal process in a network environment, Environmental Earth Sciences 74(10) (2015), 7133–7146.

11.

Zhang

et al., Quality of service control in multiple hop wireless communication environments, US, US 8000243 B2 (2011).

12.

Pandya

Bhatt

and Andharia

, Bayes estimation of two-phase linear regression model, International Journal of Quality Statistics & Reliability 2011 (2011).

13.

and Wang

, Approaches based on 2-tuple linguistic power aggregation operators for multiple attribute group decision making under linguistic environment, Elsevier Science Publishers (2011).

14.

Yang

X.S.

, Harmony search as a metaheuristic algorithm, Springer Berlin Heidelberg (2009), 1–14.

15.

Sun

Song

and Wang

, Energy-balanced clustering routing protocol based on task separation in wireless sensor networks, International Conference on Biomedical Engineering and Informatics, IEEE (2015), 778–782.