| Diseases and pests have always been an unavoidable problem in agricultural production.In recent years,due to the impact of climate change,industrial activities,farming system change,and cultivar replacement,the grain yield reduction caused by pests is becoming more and more frequent,and pests show a high and frequent trend.For example,the invasion of Spodoptera frugiperda in China in 2019 and the locust outbreak in East Africa in 2020.Therefore,how to improve the level of pest control in the process of agricultural production has become a major problem.As a traditional pest control method,chemical control can tackle the issue of pests in agricultural production to a certain degree,but it is harmful to human health,food safety,ecological environment,and bio-diversity.Meanwhile,the extensive use of chemical pesticides will lead to the rapid development of pest resistance as well.Because of the advantages of low cost,friendly environment and low side effects,solar insecticidal lamps(SILs),a green physical control method,have been widely used for pest control.Through combining SILs with Wireless Sensor Networks(WSNs),a novel agricultural Internet of Things(Io T),referred to as SIL-Io T,is established,in which SILs can communicate with each other,and also with other heterogeneous sensor nodes in farmland.SIL-Io T can complete not only real-time monitoring of the pest status,but also remote monitoring of the working status of various heterogeneous equipment.In addition,each node in SIL-Io T can estimate the number of killed insects by counting the discharge times and transmit the pest information to users,e.g.,farmers and plant protection personnel,so as to help them make reasonable decisions on the schedule and quantity of chemical control.As the fundamental problems in SIL-Io T,the deployment of SIL nodes has a direct impact on the effectiveness of routing and data fusion operations as well as on the accuracy of anticipated coverage in several agricultural scenarios,e.g.,mono-crop and mixed-crop farmlands.However,considering the complex geomorphic characteristics of actual farmland,assumptions in existing node deployment methods conflict with the actual deployment environment,so these deployment methods cannot be directly applied to solve the node deployment problem in SIL-Io T.For example,due to the existence of ridges,the actual farmland has a natural regional structure,and the boundary is always irregular,but for most existing node deployment methods,they assume that the monitoring area is complete and regular.As there are some obstacles such as hills and pools,the leeward region and the region close to pools are preferred areas for deploying SILs since these areas are more prone to pests,but almost the existing node deployment methods assume that the monitoring intensity is uniform for complete region.Therefore,in this paper,the optimal deployment problem in SIL-Io T is studied from the perspective of minimizing the deployment cost without violating the requirements of complete coverage and connectivity,and the critical contributions of this paper are summarized as follows:(1)Considering that the effective modeling of the actual farmland with partition structure is the key to carry out the research of SIL node deployment strategy,most of analogue maps in network simulators cannot effectively reflect the natural regional structure of actual farmland.As a result,the node locations obtained by the deployment algorithms in these analogue maps cannot be mapped to the actual farmland,especially for nodes with position constraints.To solve this problem,a methodology of constructing analogue maps with the characteristic of partition structure to assist the investigation of node deployment algorithms in agricultural scenarios is presented.Without loss generality,a set of vertices simulating the morphological characteristics of ridges is generated,and the relative neighborhood graph is used to construct planar neighborhood graph.Secondly,the vertices with node degree 1 in planar neighborhood graph are iteratively deleted.Thirdly,a set of the smallest rings in planar neighborhood graph is determined by the depth first search algorithm.Finally,with the aid of the offset algorithm for closed 2D lines with islands,the sub-ring of each smallest ring in planar neighborhood graph is determined.(2)Since the area coverage problem is a high-dimensional combinatorial problem,the node deployment algorithm applied in industrial Internet of Things cannot quickly converge to a high-quality solution due to the limitation of the complex geomorphic characteristic of actual farmland.In consideration of partition structure caused by natural geomorphic features,namely the actual farmland is divided into many subareas with various shapes by ridges,the problem of fully covering the whole farmland can be transformed into a set of sub-problems,i.e.,the full coverage for each subarea,so that the high-dimensional solution vector is split into many segments,and each segment contains a low-dimensional solution vector to a sub-problem of the original problem.Taking the advantage of partition structure,a partition-based node deployment strategy is proposed for mono-crop farmland,which consists of two deployment methods.These two deployment methods are same in optimization objectives,but different in deployment sequence.With the aid of these two deployment methods,the deployment cost can be minimized without violating the requirements of complete coverage and connectivity.(3)To reduce the impact on agricultural machinery services,there are some physical constraints on the geographic locations of SIL nodes.By using the candidate locations on the ridge,the constraint enforced by special requirements on the locations of SIL nodes is approximated.In other words,by adopting a discrete optimization problem,a non-convex continuous optimization problem is approximated.Since the phototaxis of pests is different not only in species,but also in geography and seasons,various kinds of pests have different requirements on the effective killing distance.Moreover,due to the fact that the canopy height,leaf area and planting density of different crops are various,the path loss of wireless signal in different mono-crop plots is various as well.Therefore,the node deployment environment in mixed-crop farmland can be regarded as a heterogeneous deployment environment.Based on greedy algorithm,a hole aware node deployment method is proposed for solving the SIL node deployment problem in mixed-crop farmland.With the aid of the proposed deployment method,the optimal deployment of heterogeneous SIL nodes in mixed-crop farmland can be realized.(4)Considering the various types of crops planted in mixed-crop farmland,the coverage intensity of different mono-crop plots in actual agricultural application is also different.The plots planted with high economic value crops should have greater coverage intensity than those planted with low economic value crops.Because of the difference of economic value of crops in mixed-crop farmland,only minimizing the deployment cost is far from the formulation of an actual agricultural application.By assigning weight to each candidate location,and jointly optimizing the deployment by maximizing the total weight and minimizing the deployment cost can effectively model the actual agricultural application.Therefore,a layered deployment method based on greedy algorithm is presented.With the help of the presented deployment method,the tendentious deployment of SIL nodes in mixed-crop farmland can be realized without violating the requirements of complete coverage and connectivity.(5)By integrating user management,farmland image management,database,deployment diagram management and other subsystems,combined with manual input information,the prototype system for SIL-Io T deployment diagram is developed to achieve the SIL node deployment locations that automatically generated. |
PDF Full Text Request