Self-Organization in Autonomous Sensor/Actuator Networks [SelfOrg]

Transkript

1 Self-Organization in Autonomous Sensor/Actuator Networks [SelfOrg] Dr.-Ing. Falko Dressler Computer Networks and Communication Systems Department of Computer Sciences University of Erlangen-Nürnberg [SelfOrg], SS

2 Overview Self-Organization Basic methodologies of self-organization; comparison of central and hierarchical control, distributed systems, and autonomous behavior; examples of self-organization Mobile Sensor/Actuator Networks Ad hoc routing; reliable communication and congestion control; sensor assistance for mobile robots; applications Coordination of Autonomous Systems Coordination and synchronization; communication aspects; clustering Bio-inspired Mechanisms Swarm intelligence; artificial immune system; intra/inter cellular information exchange [SelfOrg], SS

3 Energy Management Battery management schemes Transmission power management System power management [SelfOrg], SS

4 Energy Management Motivation Limited energy reserve Remember the primary reason for ad hoc networks: communication without fixed network infrastructure Increasing gap between power consumption and power availability Difficulties in replacing the batteries Application scenarios include hazardous environments, maybe even environments inaccessible for humans Lack of central coordination Inherent multi-hop routing Sometimes overhead through self-organization mechanisms [SelfOrg], SS

5 Energy Management Motivation Constraints on the battery source Battery size is indirect proportional to its capacity Selection of optimal transmission power Energy consumption increases with an increase in the transmission power (which is also a function of the distance between communicating nodes) Optimal transmission power decreases the interference among nodes, which, in turn, increases the number of simultaneous transmissions Channel utilization As seen before, a reduction of the transmission power increases frequency reuse better channel utilization Power control becomes especially important in CDMA-based systems [SelfOrg], SS

6 Classification of Energy Management Schemes Battery management schemes Transmission power management schemes System power management schemes Device management schemes Processor power management schemes Network Management Application layer Transport layer Network layer MAC layer Physical layer Energy Management [SelfOrg], SS

7 Battery Management Exploiting the inherent property of batteries to recover their charge when kept idle efficiency of mobile wireless communication can be enhanced through the use of improved battery management techniques Battery technologies Nickel-cadmium, lithium ion, nickel metal-hydride, reusable alkaline, lithium polymer Main criteria Energy density, cycle life, environmental impact, safety, cost, available supply voltage, charge/discharge characteristics Battery measures Theoretical capacity Nominal capacity Actual capacity [SelfOrg], SS

8 Battery Management Performance measures of a cell s discharge Discharge time Specific power (energy) [watt-hour-per-kilogram] Discharge current Constant current discharge Pulsed current discharge Recovery capacity effect In idle conditions, the charge of the cell recovers by increasing the idle time the theoretical capacity of the cell may be used [SelfOrg], SS

9 Battery Characteristics Battery models Depict the characteristics of the batteries Used for approximations of the remaining capacity Supply voltage scaling providing a balance between battery charge consumption and performance Battery-aware task scheduling optimization of the discharge power of the batteries; task graph + battery model optimized task schedule Dynamic power management graceful degradation of the quality of service of a transmission (if applicable for the particular application, e.g. audio/video transmission) Battery scheduling Scheduling between multiple available batteries in a mobile node Smart battery standard (SBS) Well-defined interaction between a system and its batteries [SelfOrg], SS

10 Modeling and Shaping Battery Discharge Patterns Effect of battery pulsed discharge If pulse discharge is applied to a cell, significant improvement in the specific energy delivery is realized Battery models for battery recovery depended on forthcoming transmissions: binary pulsed discharge generalized pulsed discharge Battery lifetime estimation Manufacturer-specified rated capacity, discharge plot of the battery Discharge current ratio can be computed Efficiency is calculated by the interpolation of point in the discharge plot [SelfOrg], SS

11 Binary Pulsed Discharge G=m p /N gain obtained by the scheme m p total expected number of packets transmitted N amount of charge of a fully charged battery [SelfOrg], SS

12 Battery-Scheduling Techniques Delay-free approaches As soon as a job arrives, the battery charge for processing the job will be provided from the cells without any delay Joint technique (JN) - the same amount of current is drawn equally from all the cells, i.e. each cell is discharged by 1/L of the current required Round robin technique (RR) - batteries are selected in round robin fashion, the current job gets the required energy from the selected cell Random technique (RN) - similar to RR but the cells are selected randomly RR JN [SelfOrg], SS

13 Battery-Scheduling Techniques No delay-free approaches The batteries coordinate among themselves based on their remaining charge E.g. by defining a threshold for the remaining charge all the cells which have their remaining charge greater than the threshold value become eligible for providing energy Delay-free approaches can be applied to the eligible cells Non-eligible cells stay in recovery state to maximize their capacity Further enhancements Heterogeneous battery-scheduling technique [SelfOrg], SS

14 Data Link Layer Solutions Battery-aware MAC protocol (BAMAC) Increasing the lifetime of the nodes by exploiting the recovery capacity effect of the batteries provisioning of enough idle time for the nodes by scheduling the nodes in an appropriate manner Working principle Usage of an additional table containing the remaining battery charge of each neighbor Dissemination of this information in each RTS, CTS, Data, and ACK packet Back-off mechanism providing a near round robin scheduling [SelfOrg], SS

15 Data Link Layer Solutions Battery-aware MAC protocol (BAMAC) back-off = Uniform[0, (2 n x CW min )-1] x rank x (T SIFS + T DIFS + T t ) CW min minimum size of the contention window rank position of that entry in the battery table T SIFS short IFS (inter-frame spacing) T DIFS DCF (distributed coordination function) IFS T t longest possible time to transmit a packet successfully (including RTS-CTS-Data-ACK handshake) Uniform[0, (2 n x CW min )-1] random number distributed uniformly, where n is the number of transmission attempts make so far for a packet [SelfOrg], SS

16 Data Link Layer Solutions Battery-aware MAC protocol (BAMAC) Whenever a node gains access to the channel, it is allowed to transmit only one packet, giving rise to an average idle time of (N-1) x m, where N is the total number of nodes contending for the common channel and m is the average time taken by a node for transmission of a packet [SelfOrg], SS

17 Network Layer Solutions Traffic-shaping schemes Main idea: introduction of delay slots in the battery discharge process Example: blocking relay traffic Random strategy: when the k th node in the path (route r) receives a session request, it accepts it with a probability sympathy(k,r) Pay-for-it strategy: each node keeps an account the help it had received form others nodes relaying its messages (credit), and the amount of help it has given to others by allowing the relay traffic (debit) each node tries to balance credit and debit Other approaches Battery energy-efficient routing protocol Shaping + scheduling Energy conserving routing based on battery status Focus on battery recovery [SelfOrg], SS

18 Classification of Energy Management Schemes Battery management schemes Transmission power management schemes System power management schemes Device management schemes Processor power management schemes Network Management Application layer Transport layer Network layer MAC layer Physical layer Energy Management [SelfOrg], SS

19 Transmission Power Management Schemes Major component in energy consumption: communications module Transmission of data Idle-listening Variation in transmission power great influence in reachability of a node Increasing the transmission range increased coverage increased power consumption Optimization of the trade-off [SelfOrg], SS

20 Data Link Layer Solutions Dynamic power adjustment based on the link affinity Frequent link failures due to node mobility reduced throughput Introduction of the stability of a route: affinity a nm high if δsnm( avg ) > 0 anm = Sthresh Snm( current) S otherwise thresh δs δs nm(avg) nm( avg ) Periodic exchange of Hello packets with constant power Calculation of the signal strength of the transmitter (S t,t+τ ) S t, t + τ S = S S H H S thresh H S a thresh * τ if moving farther andτ < a if moving closer andτ < a otherwise Calculation of the adjusted transmission power (P t,t+τ ) Sthresh Pt, t + τ = PT * St, t + τ S nm(current) current signal strength given threshold for the signal strength average of the rate of change of signal strength [SelfOrg], SS

21 Network Layer Solutions Common power protocol (COMPOW) Optimization of the transmit power level side effect: minimum interference in the network High power level interference among nodes Low power level reachability between nodes may become very low [SelfOrg], SS

22 Network Layer Solutions Common power protocol (COMPOW) Two successful simultaneous transmissions if: R R r + r 2 Major weakness: in practical wireless environments, the number of nodes and the area of the domain may not be known a priori. In such cases, a feedback strategy can be employed. [SelfOrg], SS

23 Network Layer Solutions Other solutions Globalized power-aware routing techniques Minimum power consumption routing Minimum battery cost routing Localized power-aware routing techniques Power-/Cost-/Power-cost-saving localized routing algorithm Energy-efficient ad hoc wireless network design algorithm Covering Reconfiguration [SelfOrg], SS

24 Classification of Energy Management Schemes Battery management schemes Transmission power management schemes System power management schemes Device management schemes Processor power management schemes Network Management Application layer Transport layer Network layer MAC layer Physical layer Energy Management [SelfOrg], SS

25 Energy Consumption A back of the envelope estimation Number of instructions Energy per instruction: 1 nj Small battery ( smart dust ): 1 J = 1 Ws Corresponds: 10 9 instructions! Lifetime Or: Require a single day operational lifetime = 24x60x60 =86400 s 1 Ws / 86400s 11.5 μw as max. sustained power consumption! Not feasible! [SelfOrg], SS

26 Multiple Power Consumption Modes Way out: Do not run sensor node at full operation all the time If nothing to do, switch to power safe mode Question: When to throttle down? How to wake up again? Typical modes Controller: Active, idle, sleep Radio mode: Turn on/off transmitter/receiver, both Multiple modes possible, deeper sleep modes Strongly depends on hardware TI MSP MHz, 3V): Fully operation 1.2 mw Deepest sleep mode 0.3 μw only woken up by external interrupts (not even timer is running any more) Atmel ATMega Operational mode: 15 mw active, 6 mw idle Sleep mode: 75 μw [SelfOrg], SS

27 Switching Between Modes Simplest idea: Greedily switch to lower mode whenever possible Problem: Time and power consumption required to reach higher modes not negligible Switching only pays off if E saved > E overhead Time overhead must be considered Example: Event-triggered wake up from sleep mode E saved E overhead Scheduling problem with uncertainty P active P sleep t 1 τ down t event τ up time [SelfOrg], SS

28 Alternative: Dynamic Voltage Scaling Switching modes complicated by uncertainty how long a sleep time is available Alternative: Low supply voltage & clock Dynamic voltage scaling (DVS) Rationale: Power consumption P depends on Clock frequency Square of supply voltage P / f V 2 Lower clock allows lower supply voltage Easy to switch to higher clock But: execution takes longer

29 Processor Power Management Schemes Power-saving modes Key idea: remain in sleep mode as long as possible Example: RAS remote activated switch Receiver and control logic can be turned off until a packet is received Caution: the preamble must be long enough for turning on and initializing the receiver [SelfOrg], SS

30 Transmitter Power/Energy Consumption for n Bits Amplifier power: P amp = α amp + β amp P tx P tx radiated power α amp, β amp constants depending on model Highest efficiency (η = P tx / P amp ) at maximum output power In addition: transmitter electronics needs power P txelec Time to transmit n bits: n / (R x R code ) R nomial data rate, R code coding rate To leave sleep mode Time T start, average power P start E tx = T start P start + n / (R x R code ) (P txelec + α amp + β amp P tx ) Simplification: Modulation not considered [SelfOrg], SS

31 Computation vs. Communication Energy Cost Tradeoff? Directly comparing computation/communication energy cost not possible But: put them into perspective! Energy ratio of sending one bit vs. computing one instruction : Anything between 220 and 2900 in the literature To communicate (send & receive) one kilobyte = computing three million instructions! Hence: try to compute instead of communicate whenever possible Key technique in WSN in-network processing! Exploit compression schemes, intelligent coding schemes, [SelfOrg], SS