From Distributed Software Systems Group, University of Massachusetts, Boston

BiSNET: Biologically-inspired architecture for Sensor NETworks

"One for all, all for one"
Athos, The Man in the Iron Mask

Executive Summary:

This project investigates the autonomy, adaptability, self-healing and simplicity of multi-model wireless sensor networks (MWSNs) by applying a set of biologically-inspired mechanisms.

Objectives:

This project investigates four key challenges that MWSNs face now: autonomy, adaptability, self-healing and simplicity.

1. Autonomy: MWSNs are expected to operate in an unattended area (e.g., in a forest or at sea) or physically unreachable area (e.g., inside building infrastructure). Thus, sensor nodes are required to operate with the minimum aid from base stations or human administrators.
2. Adaptability: MWSNs are required to operate through adapting to the environmental changes that sensors monitor. For example, sensor nodes may increase their sleep periods when there is no significant changes in their sensor readings. This results in less power consumption in the nodes. Also, when neighboring nodes report environmental changes, a node may draw inference from the reportsand decrease its sleep period to be more watchful for a potential local environmental change in the future. This increases the responsiveness of the node to transmit its sensor data to a base station. In addition, a node may aggregate data from different types of neighboring nodes (e.g., temperature and CO data) and transmit the aggregated data to a base station. This can reduce power consumption in the nodes on the path toward the base station.
3. Self-healing: Sensor reading usually contains some noises; it may be a false positive due to, for example, malfunction or miscalibration of sensors. Sensor nodes are required to collectively self-heal (i.e., detect and eliminate) false positives in their sensor readings instead of transmitting them to base stations. This can reduce power consumption of nodes because in-node data processing incurs much less power consumption than data transmission does.
4. Simplicity: Sensor control software needs to be simple in its design and lightweight in its implementation because of limited availability of CPU power, memory and battery.

Approach:

In order to address to the above four challenges, this project proposes an architecture for MWSN applications, called BiSNET (Biologically-inspired architecture for Sensor NETworks). BiSNET is motivated by the observation that various biological systems have already developed the mechanisms necessary to overcome those challenges. For example, bees act autonomously, influenced by local conditions and local interactions with other bees. A bee colony adapts to dynamic environmental conditions. When the amount of honey in a hive is low, many bees leave the hive to gather nectar from flowers. When the hive is full of honey, bees expand the hive. Also, bees recover (or self-heal) their pheromone traces to flowers when a part of them is lost. The structure and behavior of each bee are very simple. Based on this observation, we believe that, if MWSN applications are designed after certain biological principles and mechanisms, they may be able to meet the challenges described above (i.e., autonomy, adaptability, self-healing and simplicity).

The BiSNET runtime operates atop of TinyOS in each sensor node. It consists of a middleware platform and one or more agents, which are modeled as flowers and bees, respectively. Each MWSN application is designed as a decentralized collection of multiple agents. This is analogous to a bee colony (application) consisting of multiple bees (agents). They collect sensor data (nectars) on platforms (flowers), and carry the data to base stations, which are modeled as nests for bees, using biological behaviors such as phermone emission, replication, migration and death. A middleware platform runs on each sensor node, and hosts one or more agents. It controls the operation of the underlying node (e.g., sleep, listen and broadcast states), and provides runtime services that agents use to perform their functionalities and behaviors

Agents are designed to follow the following four design principles.

1. Decentralization: Similar to biological systems (e.g., bee colony), there are no centralized entities in BiSNET to control and coordinate agents. Decentralization allows agents to be scalable and simple by avoiding a single point of performance bottlenecks and failures and by avoiding any central coordination in deploying agents.
2. Autonomy: Similar to biological entities (e.g., bees), agents sense their local environments, and based on the sensed conditions, they autonomously behave without any intervention from/to other agents, base stations and human administrators.
3. Food gathering and storage: Biological entities strive to seek and consume food for living. For example, each bees gather nectar from flowers and digest it to produce honey. In BiSNET, each agent (bee) reads sensor data (nectar) on a platform (flower) in each duty cycle, and digest it to energy (honey). (Energy gain is proportional to a absolute change between the current and previous sensor readings.) They keep some of the energy and deposit the rest in the local platform.
4. Natural selection: The abundance or scarcity of stored energy in agents affects their behaviors and triggers natural selection. For example, an energy abundance indicates a significant change in sensor reading; thus, an agent replicates itself, and the replicated agent migrates to a neighboring node for reporting sensor data to a base station. An energy scarcity (an indication of few change in sensor reading) causes the death of agents. Like in biological natural selection where more favorable species in an environment becomes more abundant, the population of agents dynamically changes based on their energy levels (i.e., changes in their sensor readings).

Each agent implements a set of biological behaviors.

• Pheromone emission: Agents may emit different types of pheromones (replication pheromones and migration pheromones) according to their local and surrounding network conditions. Agents emit replication pheromones in response to the abundance of stored energy (i.e., significant changes in their sensor readings). Different types of agents emit different types of replication pheromones, each of which carries sensor data. For example, temperature sensing agents emit temperature pheromones, which carry temperature data. CO sensing agents emit CO pheromones, which carry CO data. Replication pheromones stimulate the agents on the local and neighboring nodes to replicate themselves. Each replication pheromone can spread to one-hop away neighboring nodes. On the other hand, agents emit migration pheromones on their local nodes when they migrate to neighboring nodes. Each replication and migration pheromone has its own concentration (or strength). The concentration gradually decays at each duty cycle. A pheromone disappears when its concentration becomes zero.
• Replication: Agents may make a copy of themselves. Each EA initiates replication only when its energy level is high and enough types of strong pheromones are available on the local node. For example, a temperature sensing agent may be configured to replicate itself only when both temperature and CO pheromones are available on the local node. Each child agent aggregates multiple sensor data stored in multiple types of available pheromones and is intended to move toward a base station to report aggregated sensor data.
• Migration: Agents may move from one node (platform) to another in response to energy abundance (i.e., indication of a significant change in sensor reading). Each agent may implement one of or a combination of the following four migration policies:

• Directional walk: Each agent may move to the nearest base station through the shortest path. Each base station periodically propagates base station pheromones, whose concentration decays on a hop-by-hop basis. Using base station pheromones, agents can sense where base stations exist approximately, and move toward the base stations by climbing pheromone gradients.
• Chemotaxis: Agents may move to base stations by following migration pheromone traces on which many other agents travel. These traces can be the shortest paths to the base stations. When there are no migration pheromones on neighboring nodes, agents perform directional walk.
• Detour walk: Each agent may go off a migration pheromone trace and follows another path to a base station when the concentration of migration pheromones is too high on the trace (i.e., when too many agents follow the same migration path). The detour walk serves two main purposes:
  • This avoids separating the network into islands. The network can be separated with the migration paths that too many agents follow, because the nodes on the paths consume more power than others and they go down earlier than others.
  • In addition to the detour with migration pheromones, agents may avoid moving through the nodes where the concentration of replication pheromones is too high (i.e., where agents detect significant changes in their sensor readings). This detour walk distributes power consumption of agent migration over the nodes where agents do not detect no changes in their sensor readings, thereby avoiding the network to be separated.

• Energy exchange: Agents always share their energy units (honey) with each other on a platform so that their energy levels become equal. A migrating agent shares its energy units with other agents on a destination platform. Also, agents periodically deposit some of their energy units to their local platforms.
• Death: Agents die due to energy starvation when they cannot balance energy gain and expenditure. The death behavior is intended to eliminate agents that carry false positive sensor data.

Novelty:

This project conveys the following contributions to the design and implementation of MWSN applications.

• Adaptive and decentralized duty cycle management: BiSNET is the first attempt to investigate dynamic duty cycle management that adaptively balances the tradeoff between power efficiency and sensing responsiveness for potential environmental changes. Each sensor node (platform) autonomously adjusts its duty cycle in a decentralized manner. The existing duty cycle management focuses on improving power efficiency; however, they do not address sensing responsiveness for potential environmental changes (i.e., the risk to miss significant environmental changes during sleep periods).
• Adaptive and decentralized data aggregation and data transmisison: . In BiSNET, agents aggregate data from different types of nodes (e.g., temperature and CO data) and transmit the aggregated data to base stations. Agents also adaptively vary their migration paths to a base station when they transmit sensor data to the base station very often. The data aggregation and migration path adjustment can reduce power consumption in the nodes on the paths toward base stations. This avoids a network to be separated into islands by the migration paths on which too many agents travel.
• Decentralized and collaborative filtering of false positive sensor data: Agents self-heal (i.e., detect and eliminate) false positive data in a decentralized manner. BiSNET provides two self-healing capabilities: intra-node self-healing and inter-node self-healing. Intra-node self-healing allows each agent to gradually suppress the emission of false positive data (pheromones) and eventually stop sending them to its neighboring nodes. Inter-node self-healing allows each agent to gradually filter out incoming false positive data (pheromones) from a neighboring node and eventually stop replicating itself with the pheromones. Both types of self-healing are achieved with the same mechanism; as a result of agents to make decisions on whether they replicate or die.
• Simple yet powerful and scalable architectural design: BiSNET implements a set of simple biological concepts and mechanisms. This contributes for the BiSNET runtime to be lightweight (1.0 KB RAM and 24 ROM in a MICA2 mote). Despite the simplicity, BiSNET exhibits powerful intelligence in MWSNs such as adaptive data transmission, inferencing, data aggregation, adaptive duty cycle management and self-healing of false positive data. Also, BiSNET allows agents to scale and retain their power efficiency against the increase of network size and transmitted data volume.

Publications:

• P. Boonma and J. Suzuki, "BiSNET: A Biologically-Inspired Middleware Architecture for Self-Managing Wireless Sensor Networks," In Elsevier Journal of Computer Networks, Special Issue on Middleware Challenges for Next Generation Networks and Services, 51(16), pp. 4599 - 4616, November 2007. (33% acceptance rate)

• P. Boonma and J. Suzuki, "An Adaptive, Scalable and Self-Healing Sensor Network Architecture for Autonomous Coastal Environmental Monitoring," In Proc. of the 6th IEEE Conference on Technologies For Homeland Security, June 2007.

• P. Boonma and J. Suzuki, "A Biologically-Inspired Sensor Network Architecture for Autonomous Ecological Observation," In Proc. of the 5th Elsevier/ISEI International Conference on Ecological Informatics (ISEI5), Santa Barbara, CA, December 2006.

• P. Boonma and J. Suzuki, "Biologically-Inspired Data Aggregation for Multi-Modal Wireless Sensor Networks," In Proc. of the 31st IEEE Conference on Local Computer Networks (IEEE LCN), Tampa, FL, November 2006. (34% acceptance rate)

• P. Boonma and J. Suzuki, "BIGBAND: A Novel Software Framework for Ecological Wireless Sensor Networks," In Proc. of 5th Annual New England Environmental Research Symposium, poster paper, Bridgewater, MA, November 2006.

• P. Boonma, P. Champrasert and J. Suzuki, "A Biologically-Inspired Architecture for Self-Managing Sensor Networks," In Proc. of the 3rd IEEE Confernece on Sensor, Mesh and Ad Hoc Communications and Networks (SECON), IWWAN subtrack, Reston, VA, September 2006.

• P. Boonma, P. Champrasert and J. Suzuki, "BiSNET: A Biologically-Inspired Architecture for Wireless Sensor Networks," In Proc. of the 2nd IEEE/IARIA International Conference on Autonomic and Autonomous Systems (IEEE/IARIA ICAS), Santa Clara, CA, July 2006.