Skip to main content

Acoustic Actuated Sensor Networks for Industrial Processes (AASN4IP)

1st September 2008 to 29th February 2012

The aim of this project is to develop an underwater mobile sensor network for exploring and monitoring enclosed and cluttered underwater environments like nuclear waste storage ponds. Nuclear power provides a significant portion of our energy demands and is likely to become more wide spread with the growing world population. However, the radioactive waste generated in these power plants must be held for 60-100 years underwater in a storage pool in large metal containers. These underwater environments must be carefully monitored and controlled to avoid an environmental catastrophe. Further details on the motivation and aims of our project can be found 

here. In addition to the nuclear waste storage, there is a wide range of applications that can also benefit from our underwater mobile sensor network. One of these applications is industrial scale chemical process monitoring. Generally the chemical reactions are carried out in large reaction vessels and the chemical engineers are interested in monitoring the conditions inside the reaction vessel. Our small scale mobile sensor nodes can be introduced in such vessels to provide in situ monitoring. Another interesting application is the monitoring of a network of water reservoirs and pipes in a waste water treatment plant.

The mobile sensor nodes that we envisage for these tasks are much smaller in size as compared to the traditional underwater vehicles. Our mobile sensor nodes are just 10cm in diameter and have full six degrees of freedom. This allows them to easily move in a cluttered environment. These robots are not only equipped with various sensors to sense the environment but also have ultrasound transducers for inter-node communication and range measurements. The following photograph shows one of the prototypes of our mobile sensor node.


Robot Prototype
Robot Prototype
 Robot Prototype (Photos from Simon Watson, University of Manchester)


For the robots to sense and explore the storage pools, they must be  able to determine their positions in these underwater environments.  However, the cluttered nature of our application scenarios presents us  with unique challenges. The ultrasound pulses used for performing  range measurements can reflect and bounce off multiple surfaces before  arriving at the the transducer. These multipath reflections introduce  large positive errors in some of the estimated distances between the  sensor nodes. These erroneous measurements make it very difficult to  estimate the true positions of sensor nodes. As part of the AASN4IP  project, we have developed two different approaches for solve these  problems.

Our first approach involves a single-hop scenario where the mobile  sensor nodes perform range measurements to fixed reference points or  anchor nodes. These are high powered ultrasound transducers attached  to the pool infrastructure or vessel walls at known coordinates. The  mobile sensor nodes use the range measurements to these anchor nodes  and their coordinates to estimate their own coordinates. Since some of  these range measurements are erroneous, the traditional approaches  that employ the least squares technique fail to provide an accurate  position estimate when all the measurements are used to estimate node  coordinates. We propose an algorithm based on minimizing the sum of  absolute values of the residuals for these scenarios and show with  real experiments that it performs significantly better than the  traditional approaches. Our algorithm is independent of the underlying  physical layer used to perform range measurements and thus can be used  with any of the newer radio technologies e.g. UWB as well.  Please see  our publication for further details of this work .

Our second approach involves a multi-hop scenario where  special-purpose nodes, called 'localizers', are deployed in cluttered  NLOS-prone environments to help localize sensor nodes performing the  monitoring task. We investigate the performance of distributed  localization techniques, such as iterative localization and  DV-Distance, in cluttered environments, and explore a variety of  scenarios in which they outperform one-hop localization. The key  observation from our initial study is that the accuracy of distributed  localization techniques largely depends on the placement of  localizers. This advocates the need for careful placement of  localizers in a cluttered environment to minimize localization error.  In our initial work, we have focused on DV-Distance, and proposed an  algorithm that carefully selects where localizers should be placed to  reduce the localization error of DV-Distance. Our proposed algorithms  is centralized, it assumes knowledge of the clutter topology, and does  not deal with the practical problem of actually moving the localizers  to their selected positions. In order to address these problems, we  are currently working on distributed algorithms that enable a swarm of  localizers to coordinate with each other and self-deploy to provide a  high-accuracy localization service for the sensor nodes.  Please see  our publication for further details of this work


Selected Publications

View All


Principal Investigator


Share this: