Lessons in Implementing Bio-inspired Algorithms on Wireless Sensor Networks

Abstract

The demand for highly lightweight decentralised selfmanagement of Wireless Sensor Networks has lead to the pursuit of emergent or bio-inspired solutions. However, many of the algorithms produced to manage a WSN focus' on one managerial aspect or parameter limiting their usefulness and consuming already scarce resources. We have identified sets of common structures and elements of many well-known emergent algorithms. In this paper present one example algorithm that exploited this knowledge to efficiently manage more than one managerial parameter. This algorithm was then tested using simulations (a standard practice for the field). However, when implementing the algorithm on actual devices we soon found some unexpected results. We discuss this phenomenon and suggest some causes aiming to illustrate that current WSN bio-inspired research simulations have limited usefulness in the real world. Node synchronization while disseminating data over WSNsWireless sensor networks (WSN) are networks of small resource constrained devices which sense the environment and report the results via wireless networks. They allow spatial or temporal measurements of phenomenon previously difficult to analyse [3]. One of the current challenges in the WSN field is the development of management systems which allow WSN to be easily deployed in various application domains [4,6] as different WSN application domains often have different management requirements. However after our five years experience in developing algorithms and applications for WSN we have began to see many common requirements to all. One such requirement is that global system parameters such as time and location need to be managed and ideally together to save resources. However given that WSN are very restricted in terms of resources and usually battery powered, overheads involving communication are to be avoided as much as possible.Therefore lightweight decentralised management solutions are favoured. To this end, we have been investigating the usage of Bio-inspired solutions to the maintenance of global system parameters where the management solution emerges.A definition of an emergent algorithm is given in [2] which state that an emergent algorithm produces predictable, or stable global effects with respect to one or more system parameters, by communicating with only immediate neighbours and in the absence of global control or information. Our method of engineering emergent algorithms is to base them on a simple algorithmic structure which is common to many algorithms that show emergent behaviour.Many bio-inspired algorithms use simple rules and local information to create a global consensus of a single parameter, such as clock time in the flashing firefly algorithm [7], or location in the slime algorithm [11]. Our research has identified the control loop as the core structure of several bio-inspired algorithms and within this we can combine loops either serially, and nested to allow the control of multiple parameters on a global level without the use of a centralised poin...

A Systematic Review on Firefly Algorithm: Past, Present, and Future

“…Breza et al [98] used nature inspired solution for lightweight WSN decentralization. They used few bio inspired parameter for efficient management of WSN.…”

“…he used FA to do this optimization based on local search and find it efficient. The list of various engineering application shown in table4 [83] Rigid image registration Image Processing [84] Image vector quantization Image Processing [85] Vector quantization for image compression Image Processing [86] Non-linear grayscale image enhancement Image Processing [87] Image segmentation Image Processing [88] Multilevel thresholding Image Processing [89] Multilevel image thresholding selection Image Processing [90] Cross entropy threshold Image Processing [91] Linear array antenna Antenna Designing [92] 2 ring circular array antenna Antenna Designing [93] Linear antenna designing Antenna Designing [94] Concentric ring array antenna Antenna Designing [95] Self-synchronization of robot Robotics [96] Industrial robots Robotics [97] Optimal semantic web service composition Semantic Web [98] Wireless sensor networks Wireless Network [99] Support vector machine Business Optimization [100] Financial portfolio optimization Business Optimization [101] Co-variance matrix adaptation Chemical Engineering [102] Tower structures Civil Engineering [103] Optimum design of structures Civil Engineering [104] Optimum design of trusses Civil Engineering [105] Precipitation field Meteorology Optimization [106] Economic emissions load dispatch problem Industry Optimization [107] Non-convex economic dispatch problems Industry Optimization [108] Steel slabs Casting Industry Optimization [109] Ring array antenna Industry Optimization [110] Simulated manufacturing process Industry Optimization [111] Unit commitment Industry Optimization [112] Demand response scheduling Industry Optimization [113] Train's energy saving Industry Optimization…”

Formal Verification of Synchronisation, Gossip and Environmental Effects for Critical IoT Systems

“…This kind of property is said to specify the probability that the model is in thè idle' state inntely often. FiGo is typical of a class of algorithms that combine rey synchronisation [42] and gossip protocols [21] into a single epidemic process [9]. This mix- Note that s1Phase refers to the phase of the rst sensor node module, which is called s1.…”

Cost-efficient deployment of multi-hop wireless networks over disaster areas using multi-objective meta-heuristics

“…Notwithstanding the flurry of research around nature-inspired optimization (and in general, bio-inspired processing) for wireless network applications, it is a common belief [69,70] that as many techniques as possible should be explored towards overcoming the large gap from the simulation of this kind of methods to their implementation in practical systems. This observation is indeed where the contribution of our research work is framed: to the best of our knowledge the deployment of multi-hop wireless networks under a multiobjective cost-coverage criterion and redundancy constraints has not been approached yet using HS-based heuristics nor tested in realistic simulation setups, hence advancing over recent contributions using this algorithm that focus instead on maximizing the energy efficiency of fixed networks [67,68].…”