Connectivity-Aware Pheromone Mobility Model for Autonomous UA V Networks Shreyas Devaraju Alexander Ihlery and Sunil Kumarz

2025-04-24 0 0 601.16KB 7 页 10玖币

侵权投诉

Connectivity-Aware Pheromone Mobility Model for

Autonomous UAV Networks

Shreyas Devaraju∗, Alexander Ihler†, and Sunil Kumar ∗‡

∗Computational Science Research Center, San Diego State University, San Diego, CA, USA

†School of Information & Computer Science, University of California, Irvine, CA, USA

‡Electrical & Computer Engineering Department, San Diego State University, San Diego, CA, USA

Email: sdevaraju@sdsu.edu, ihler@ics.uci.edu, skumar@sdsu.edu

Abstract—UAV networks consisting of reduced size, weight,

and power (low SWaP) ﬁxed-wing UAVs are used for civilian

and military applications such as search and rescue, surveillance,

and tracking. To carry out these operations efﬁciently, there is a

need to develop scalable, decentralized autonomous UAV network

architectures with high network connectivity. However, the area

coverage and the network connectivity requirements exhibit

a fundamental trade-off. In this paper, a connectivity-aware

pheromone mobility (CAP) model is designed for search and

rescue operations, which is capable of maintaining connectivity

among UAVs in the network. We use stigmergy-based digital

pheromone maps along with distance-based local connectivity

information to autonomously coordinate the UAV movements, in

order to improve its map coverage efﬁciency while maintaining

high network connectivity.

Index Terms—Airborne network, UAV network, search and

rescue, network connectivity, pheromone model.

I. INTRODUCTION

The unmanned aerial vehicles (UAVs), equipped with self-

localization and sensing capabilities, are used in applications

such as search-and-rescue, tracking and surveillance [1]–[4].

Distributed UAV networks are scalable, sense simultaneously

in an expanded area, and do not have a single point of failure.

In distributed or decentralized, autonomous UAV network

architectures, the nodes perform only local sensing and com-

municate with their neighbors without any global knowledge

[1]. However, such networks can face communication issues,

since low SWaP (size, weight and power) UAVs have a limited

communication range. Therefore, the connectivity among the

UAVs must be maintained to allow their coordination and

control. Whereas a high network connectivity facilitates better

communication among the UAVs, an increase in coverage

performance leads to a faster discovery and better tracking

of targets in a search area. Note that the area coverage

and network connectivity requirements exhibit a trade-off,

i.e., dispersing the UAVs to improve coverage will typically

negatively impact connectivity [2], [3].

Swarm intelligence methodologies inspired by nature, such

as the social behavior of insects, birds, and ﬁsh, can be used to

solve complex problems cooperatively by using simple rules

and local interactions. One such widely used method is the

use of stigmergic digital pheromones [4], [5], which act as

the spatio-temporal potential ﬁelds that are used to coordi-

nate and control the UAV movement. In this paper, we use

the digital pheromone-based stigmergic algorithms to achieve

quick exploration of completely unknown environments. Their

decentralized nature makes them fault-tolerant and highly

scalable. However, most repulsion pheromone-based mobility

models focus only on coverage performance of UAV networks,

and ignore the connectivity.

This paper addresses the problem of achieving an efﬁ-

cient coverage of a given search area while preserving the

network connectivity in an autonomous, decentralized UAV

network. We design a UAV mobility model, which combine the

pheromone mobility model with local connectivity information

to optimize the coverage and connectivity performance. We

call the model as connectivity-aware pheromone mobility

(CAP) model. This mobility model selects a UAV path that

balances pheromone values with estimated connectivity values

at a number of potential waypoints.

Paper Organization: We ﬁrst review the existing schemes

for UAV mobility in Section II, followed by a brief overview

of the pheromone based UAV mobility model in Section III.

Then we describe our proposed CAP model in Section IV.

The simulation results are discussed in Section V, followed

by the conclusions in Section VI.

II. RELATED WORK

Several algorithms such as particle swarm optimization,

artiﬁcial bee colony and ant colony optimization (ACO) have

been proposed for control and coordination of swarms for

various search, rescue, and tracking applications [1], [6]. The

digital pheromone based mobility model has been used for

target search and other other similar tasks in UAV networks.

In digital pheromone schemes, information about the

pheromone map is communicated between agents in the net-

work through direct or indirect communication. In the direct

communication methods, each agent maintains a full or partial

pheromone map of its immediate vicinity. Updates in the

pheromone map due to deposits or withdrawals are communi-

cated only locally. In [7], distributed stigmergic coordination

of UAVs for automatic target recognition is done through

direct communication. The UAVs mark potential targets and

communicate the pheromone information to their neighbors

using a decentralized gossip mechanism.

Sauter et al. [5] is an example of indirect communication

scheme for controlling and coordinating the UAV swarm for

arXiv:2210.06684v1 [cs.NI] 13 Oct 2022

surveillance, target acquisition, and tracking. Here the coor-

dination of swarm members is based on digital pheromones

maintained in an artiﬁcial pheromone map, and a central-

ized base station (BS) is used to communicate the global

pheromone map to all the UAVs. Failure of the centralized

BS may lead to failure of the entire system.

Some schemes use a fusion of stigmergic pheromone algo-

rithm and ﬂocking behaviors to coordinate a group of UAVs

for performing decentralized target search [4], [8]. Here, the

UAVs deposit digital attract pheromones when a potential tar-

get is detected to attract UAVs in the area; Repel pheromones

are deposited when no target is found. They also follow Boids

[9] ﬂocking rules to organize the swarm for better perception

and communication for tracking the targets. An evolutionary

algorithm is used in [4] for tuning the pheromone and ﬂocking

behaviors to get an optimal performance. Shao et al. [10]

designed a navigation algorithm by using the pheromone

algorithm on top of the Olfati-Saber’s ﬂocking algorithm [11],

where leader-follower based ﬂocking is performed. The cov-

erage and network connectivity performance for a UAV group

using a random vs. pheromone guided mobility model are

compared in [2]. While the random model follows a Markov

process, the UAVs move to a low repel pheromone area in the

pheromone model. The pheromone model provides a better

coverage than the random model, but neither model show a

good connectivity performance. Messous et al. [14] address

the connectivity issue in UAV ﬂeets by weighting a UAV’s

tendency to follow its neighbor based on its connectivity, hop

count to the base station, and energy level. Similarly, dual-

pheromone clustering hybird approach (DPCHA) [13] uses

dual pheromones for target tracking and area coverage, and

the clustering to maintain stable network connectivity.

A. Review of CACOC2Model

The CACOC2model [12] uses the ACO with a chaotic

dynamical system (CACOC) [15], together with the Boids

ﬂocking model to maximize the coverage while preserving

the network connectivity. The CACOC model [15] uses the

pheromone mobility model, along with chaotic dynamics

obtained using the Rossler system, to obtain a deterministic

but unpredictable system. In CACOC, each UAV in the swarm

moves left (L), ahead (A) or right (R) based on the pheromone

values in its respective neighboring cells and the next value

(ρn) in the ﬁrst return map of the Rossler attractor (see Fig. 1

in [12]).

In CACOC2model, the Boids ﬂocking behavior [9], in-

cluding the collision avoidance, velocity matching and ﬂock

centering, is combined with CACOC to improve the network

connectivity. Here, the ﬂock centering forces the UAVs to

maintain connectivity. The model uses two forces [12] :

•ˆ

FCis a vector that gives a direction (L, R or A).

•ˆ

Fflock is a vector for the ﬂock centering force computed

with the average value of the last vector used for the

neighboring UAVs.

The normalized sum of the these two force vectors gives a

vector ˆ

Vwith a constant speed v[12] :

V=v·ˆ

FC+f·ˆ

Fflock

kˆ

FC+f·ˆ

Fflock k2(1)

In (1), frepresents the inﬂuence of ﬂocking force, which

determines the connectivity among the UAVs.

III. OVERVIEW OF PHEROMONE MOBILITY MODEL

The pheromone mobility model uses repel digital

pheromones to promote exploration and fast coverage of

an area with no prior information [16]. Note that a dig-

ital pheromone has the same characteristics of a natural

pheromone, such as deposition, evaporation and diffusion.

Each UAV moves towards the cells with minimum repel

pheromone value and deposits a repel pheromone of magnitude

‘1’ in the cells scanned along its trajectory. After a UAV

deposits a pheromone in a cell (x, y), it is progressively

diffused to the surrounding cells, with a constant diffusion

rate ψ∈[0,1]. This encourages UAVs to spread out and

move toward the unvisited cells. The pheromone value of each

cell also evaporates, decreasing its intensity over time by a

constant rate λ∈[0,1]. If the map environment and target

locations change with time, the evaporation of the deposited

repel pheromones over time allows for UAVs to revisit already

scanned cells of the map after a certain time gap.

For simplicity, the UAVs are assumed to move in two-

dimensional space to search a given area, which is divided

in a grid of C2cells, where each cell is identiﬁed by its (x, y)

coordinates. Pheromones deposited by each UAV in the grid

space are saved in a digital pheromone map. In a decentralized

UAV network, the UAVs exchange their digital pheromone

maps with their 1-hop neighbors by using the periodic ’hello

messages’.

Mathematically the pheromone value p(x,y)in a cell (x, y)

at time tis described as [4], [5], [8],

p(x,y)(t) = (1 −λ)·[(1 −ψ)·p(x,y)(t−1)+

∂p(x,y)(t−1, t) + ∂d(x,y)(t−1, t)] (2)

where (1 −ψ)·p(x,y)(t−1) is the pheromone value re-

maining in cell (x, y)after diffusion to the surrounding cells,

∂p(x,y)(t−1, t)is the new pheromone value deposited in

the update interval (t−1, t), and ∂d(x,y)(t−1, t)is the

additional pheromone diffused to the current cell from its eight

surrounding cells in the update interval (t−1, t), which is

described as,

∂d(x,y)(t−1, t) = ψ

8·

a=−1

b=−1

p(x+a,y+b)(t−1) (3)

IV. CONNECTIVITY-AWARE PHEROMONE MODEL

The pheromone mobility models achieve a fast coverage of

the area by pushing the UAVs away from each other. However,

this leads to poor connectivity among UAV nodes due to a

limited transmission range of UAVs. Maintaining a strong

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

Connectivity-AwarePheromoneMobilityModelforAutonomousUAVNetworksShreyasDevaraju,AlexanderIhlery,andSunilKumarzComputationalScienceResearchCenter,SanDiegoStateUniversity,SanDiego,CA,USAySchoolofInformation&ComputerScience,UniversityofCalifornia,Irvine,CA,USAzElectrical&ComputerEngineeringDepartmen...

展开>> 收起<<

Connectivity-Aware Pheromone Mobility Model for Autonomous UA V Networks Shreyas Devaraju Alexander Ihlery and Sunil Kumarz.pdf

共7页,预览2页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Connectivity-Aware Pheromone Mobility Model for Autonomous UA V Networks Shreyas Devaraju Alexander Ihlery and Sunil Kumarz

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: