260 likes | 386 Views
A Biologically-Inspired Approach to Designing Wireless Sensor Networks. Matthew Britton, Venus Shum, Lionel Sacks and Hamed Haddadi The University College London, London ,UK EWSN’04. OUTLINE. Introduction System Requirements KOS Hardware Environment Performance Analysis Conclusion.
E N D
A Biologically-Inspired Approach to DesigningWireless Sensor Networks Matthew Britton, Venus Shum, Lionel Sacks and Hamed HaddadiThe University College London, London ,UK EWSN’04
OUTLINE • Introduction • System Requirements • KOS • Hardware Environment • Performance Analysis • Conclusion
INTRODRCTION • Biological Automata have a number of desirable characteristics such as: • scalability • robustness • simplicity • self-organization • There are significant advantages in treating some classes of sensor networks as Biological automata–like system
Biological Automata Self-organise and self-optimise System adapt to dynamic environments Neighbor to neighbor interaction Iterative–like process Change slow to spread through the network INTRODRCTION (cont.) agent
INTRODRCTION(cont.) • Application to sensor network • To limit communication to short range • Avoid the centralize algorithm (power mangement) • Scalability • For environmental monitoring the size of the spatial field of interest will not be unknown in design phase • Simplicity of mangement • Self-organising and self-optimised (robust) • Dynamic environment and requirment • In environmental monitoring various temporal phases of operation
INTRODRCTION(cont.) • Iterative application • Quality of their result • Operation become simple and predictable • For relatively high-latency requirement system
INTRODRCTION(cont.) • Goal- • Decentralised management • Self-organisation and autonomy • Robustness to topological change • Limited processing power of individual nodes • Power control for individual nodes • Adaptation to dynamic environments and changing roles
System Requirements • Coordination (distributed algorithm) • Nodes within the same area interact and understand the phenomenon • Representative node coordinate other nodes action (save energy ) • “Horizontal”Layers of network function upon a network of nodes • “Vertical”tasks within one of these node
Data transport protocol Gossip-protocol Like Flooding protocol Periodically exchange state to neighbor System Requirements(cont.) C G D Select a peer D B E E C F Exchange view F G A B
System Requirements(cont.) • Power management • Cluster, avoid multihop radio communication • High integrity operation • System can adapt to failures, corrupted data or imprecision’s in parameters and still function sufficiently (Fault tolerance)
KOS Features • Modularity of application design • Simple execution model • single-tasking ,run to completion model • Highly communication oriented (messaging interface) • Power awareness • Adaptive scheduling • Simple processing load control • Adjust the execution periods of iterative app
KOS Structure • The kOS is divided into objects and methods. • Task execution is performed by specifying objects, methods and execution times Method Library routine#1 Object Main routine Method Library routine#2 Object Method
KOS Structure (cont.) • The KOS functional abstraction
KOS Operation • Task scheduling • Sleep/activity/sleep cycle • Schedule object manage transitions • Messaging handling • SAD (SECOAS APP Message Protocol) • SAM (SECOAS Data Message Protocol) • Robustness of operation
Task scheduling High priority scheduler High priority ISR High priority interrupt High priority interrupt Sleep Return to low priority ISR Low priority interrupt Low priority ISR Low priority scheduler Kos reset command Boot start WDT time-out
Task scheduling Concept(cont.) Scheduler Task Task Sleep Ready Queue Task run to completion Run ISR Preemption ISR High priority scheduler Low priority scheduler Hardware RF Sensor UART Timer
Task scheduling(cont.) • The biological automaton characteristic of iteration to design application • Scheduler can control the period of its execution. • Reduce power consumption when the node’s battery power is low. • KOS use an off-line analysis to gauge the duty cycle of each object’s iteration
Message handling • The message object is scheduled periodically after radio and sensor interface message are received • SAM is used by objects for intra- and inter-node communication (between application) • SAD is used between application and sensor module • Using message-handling services and gossip protocol disseminate information around network (policies or application parameters)
Message handling(cont.) B Periodically exchange state to neighbor A Gossip protocol D C
Message handling(cont.) • Data flow intra-node between application, radio module and sensor module SAD Radio Receive Buffer Sensor Transmit Buffer SAD SAM SAM Applications SAD Sensor Module Radio module SAM Radio Transmit Buffer Sensor Receive Buffer SAD SAM SAM
Robustness of operation • Reboot itself in an attempt to bypass any intermittent problems • WDT • Application will operate given unknown radio connectivity conditions • If information is unavailable for short periods of time, this simply halts the iterative process for that time period • Application will load-controlled by the scheduler • Change periodicity of these application
MCU: PIC18F452(8-bit 4MHz) 32K FLASH 1.5KRAM 200 bytes EEPROM Sensor module Radio module LCD display Hardware Environment
Performance Analysis • Power usage
Performance Analysis(cont.) • CPU duty cycle • 4MHz operates at 1 million instructions per second
Performance Analysis(cont.) • Memory usage
Conclusion • Treat wireless sensor networks like biological automata • Beneficial features :scalability , robustness, self-organisation • Support distributed Application