A Multiagent System Approach to Schedule Devices in Smart Homes - - PowerPoint PPT Presentation

▶
a multiagent system approach to schedule devices in smart
SMART_READER_LITE
LIVE PREVIEW

A Multiagent System Approach to Schedule Devices in Smart Homes - - PowerPoint PPT Presentation

Sep 25, 2023 153 likes •435 views

A Multiagent System Approach to Schedule Devices in Smart Homes William Yeoh Enrico Pontelli Ferdinando Fioretto New Mexico State University University of Michigan May, 2017 Home Automation 1 Network of smart homes 2 Smart Homes and SHDS

slide-1
SLIDE 1

A Multiagent System Approach to Schedule Devices in Smart Homes

Ferdinando Fioretto University of Michigan

May, 2017

William Yeoh Enrico Pontelli New Mexico State University

slide-2
SLIDE 2

1

Home Automation

slide-3
SLIDE 3

2

Network of smart homes

slide-4
SLIDE 4

Smart Home Device Scheduling (SHDS)

A SHDS problem is composed of:

  • A neighborhood of smart homes.
  • A set of smart electric devices within each home.
  • A time horizon for the device scheduling.

3 exact 2 Zi. The ener

problem uildings hi and whose Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-5
SLIDE 5

Smart Home Device Scheduling (SHDS)

A SHDS problem is composed of:

  • A neighborhood of smart homes.
  • A set of smart electric devices within each home.
  • A time horizon for the device scheduling.
  • A pricing function expressing cost per kWh of energy

consumed.

4

time start 0:00 8:00 12:00 14:00 18:00 22:00 time end 7:59 11:59 13:59 17:59 21:59 23:59 price ($) 0.198 0.225 0.249 0.849 0.225 0.198

Pacific Gas & Electric Co. pricing schema

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-6
SLIDE 6

Smart Home

A smart home has:

  • A set of smart devices it can control (e.g, HVAC, roomba)

5

exact 2 Zi. The ener

problem uildings hi and whose

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-7
SLIDE 7

Smart Home

A smart home has:

  • A set of smart devices it can control (e.g, HVAC, roomba)
  • A set of locations (e.g., living room, kitchen)

6

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-8
SLIDE 8

Smart Home

A smart home has:

  • A set of smart devices it can control (e.g, HVAC, roomba)
  • A set of locations (e.g., living room, kitchen)
  • A set of sensors (e.g., cleanliness, temperature)

7

Thermostat Cleanliness sensor Battery charge sensor

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-9
SLIDE 9

Smart Devices (Actuators)

A smart device is defined with a

  • Location: where the device can act (e.g., living room)
  • Actions it can perform (clean, charge, stop) and the power consumption

associated to them

  • Sensors’ states properties it affects (e.g., cleanliness, battery_charge)

8 Action State property Power (kW/h)

run cleanliness, battery charge 0.0 charge battery charge 0.26 stop 0.0 Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-10
SLIDE 10

Smart Devices (Sensors)

  • We associate a predictive model to each home sensor.
  • It describes the transition of a state property from a state s and

time t to time t+1, when affected by a set of actuators.

9

Thermostat

Heater Oven Current State Next State

  • ff
  • ff

12 C 11 C

  • ff

bake 12 C 13.8 C

  • n
  • ff

12 C 17.5 C

  • n

bake 12 C 19.3 C

Effect of the environment

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-11
SLIDE 11

Smart Device Schedules

Scheduling Rules

  • Active rules: specify user-defined objectives on a desired state
  • f the home. E.g.,
  • Passive rules: define implicit constraints on devices. E.g.,

10

living room cleanliness ≥ 75 before 1800

zv battery charge ≥ 0 always zv battery charge ≤ 100 always

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-12
SLIDE 12

Smart Device Schedules

Schedule: A sequence of actions for the home devices.

11

1400 1500 1600 1700 1800 15 30 45 60 75

Cleanliness (%)

15 30 45 60 75

Battery Charge (%) Time Goal Deadline

40 15 R 15 30 R 35 30 C 55 30 C 30 45 R 5 60 R 25 60 C 75 R 65 S

Device Schedule Cleanliness (%) Battery Charge (%)

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-13
SLIDE 13

min

ξ[0!H]

Zi

↵c·Csum + ↵e·Ediff

Smart Home Device Scheduling (SBDS)

  • SHDS objective:

12

Aggregated monetary cost of the homes schedules Energy consumption peaks across all homes subject to: Homes’ devices schedules

8hi 2 H, R[ta!tb]

p

2 Ri : ⇠[ta!tb]

Φp

| = R[ta!tb]

p

P

All scheduling rules must be satisfied

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-14
SLIDE 14

Distributed Constraint Optimization

13

<X, D, F, A, α>:

  • X: Set of variables.
  • D: Set of finite domains for each variable.
  • F: Set of constraints between variables.
  • A: Set of agents, controlling the variables in X.
  • α: Mapping of variables to agents.

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

exact 2 Zi. The ener

problem uildings hi and whose

slide-15
SLIDE 15

Distributed Constraint Optimization

14

<X, D, F, A, α>:

  • X: Set of variables.
  • D: Set of finite domains for each variable.
  • F: Set of constraints between variables.
  • A: Set of agents, controlling the variables in X.
  • α: Mapping of variables to agents.
  • GOAL: Find a cost minimal assignment.

x⇤ = arg max

x

F(x) = arg max

x

X

f2F

f(x|scope(f))

F

min min

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-16
SLIDE 16
  • Agents coordinate an assignment for

their variables.

  • Agents operate distributedly.

Communication:

  • By exchanging messages.
  • Restricted to agent’s local neighbors.

Knowledge:

  • Restricted to agent’s sub-problem.

15

fab fbc fac xc xb xa xd fbd

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

Distributed Constraint Optimization

slide-17
SLIDE 17

Solution Approach

SH-MGM: Adaptation of a local search DCOP algorithm (MGM).

1. Agents independently search for a feasible schedule for their local devices.

16 exact 2 Zi. The ener

problem uildings hi and whose ci : schedule cost Eit :energy consumption Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-18
SLIDE 18

exact 2 Zi. The ener

problem uildings hi and whose

Solution Approach

SH-MGM: Adaptation of a local search DCOP algorithm (MGM).

1. Agents independently search for a feasible schedule for their local devices. 2. Schedule costs and energy consumption are broadcasted to all other agents.

17

(ci , Eit ) (ci , Eit ) ci : schedule cost Eit :energy consumption Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-19
SLIDE 19

Solution Approach

SH-MGM: Adaptation of a local search DCOP algorithm (MGM).

3. Upon receiving all other agents costs and energy consumptions:

  • Computes the objective cost with its current schedule.
  • Within a time limit, it finds a new solution to its local subproblem that is no worse than

the current solution.

18

ci : schedule cost Eit :energy consumption

αc · ci(ˆ ξ[0!H]

Zi

) + αe · ˆ Ediff ≤ αc · ci(ξ[0!H]

Zi

) + αe · Ediff

current schedule new schedule Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-20
SLIDE 20

Solution Approach

SH-MGM: Adaptation of a local search DCOP algorithm (MGM).

3. Upon receiving all other agents costs and energy consumptions:

  • Computes the objective cost with its current schedule.
  • Within a time limit, it finds a new solution to its local subproblem that is no worse than

the current solution.

  • It computes the gain Gi between its current and new solutions, and broadcast it to all
  • ther agents.

19

ci : schedule cost Eit :energy consumption

Gi = ⇣ αc · ci(ξ[0!H]

Zi

) + αe · Ediff⌘ − ⇣ αc · ci(ˆ ξ[0!H]

Zi

) + αe · ˆ Ediff⌘

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-21
SLIDE 21

exact 2 Zi. The ener

problem uildings hi and whose

Solution Approach

SH-MGM: Adaptation of a local search DCOP algorithm (MGM).

  • 4. Upon receiving all other agents’ gains Gk, it checks if the agent has the largest

gain among all those received. If so, it updates its schedule to the new schedule, otherwise it retains its old schedule.

  • 5. The process repeats untill convergence (all gains = 0) or a fixed number of

iterations.

20

Gi : agent’s gain Gj Gk Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-22
SLIDE 22

Evaluation: Settings

  • 7 Raspberry Pis connected via a LAN.
  • Each controlling 9 smart actuators.
  • Communication and coordination of the MAS is implemented

via the JADE framework.

  • Each agent uses an internal CP solver (JaCoP) to solve its local

scheduling problem.

21

A Raspberry PI with a dangle Smart devices

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-23
SLIDE 23

Evaluation: Physical MAS Experiments

22

Settings:

  • H = 12 (step = 30 min)
  • Realistic device consumptions and environment settings (see our paper at

OPTMAS)

  • CP timeout = 10 sec

10 20 30 40 50 740 780 820 Number of Cycles Solution Cost

  • SH−MGM

Uncoordinated

Main Results:

  • SH-MGM finds better

solutions than a simple uncoordinated approach.

  • Solution quality increases

with the number of cycles.

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

SH-MGM vs. Uncoordinated approach

slide-24
SLIDE 24

Evaluation: Physical MAS Experiments

23

Main Results:

  • SH-MGM has better effects
  • n peaks reduction w.r.t an

uncoordinated approach.

2 4 6 8 10 10 50 200 1000

Time Total Energy Consumption (kWh)

  • uncoordinated

αc=0, αe=1 αc=0.5, αe=0.5 αc=1, αe=0

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

Settings:

  • H = 12 (step = 30 min)
  • Realistic device consumptions and environment settings (see our paper at

OPTMAS)

  • CP timeout = 10 sec

Effect of weights ⍺c and ⍺e

slide-25
SLIDE 25

Evaluation: Large scale simulation

24

Settings:

  • H = 12 (step = 30 min)
  • Realistic device consumptions and environment settings
  • CP timeout = 10 sec

City k convergence

  • avg. l.s.

network avg. cost max peak time (sec) time (sec) load ($/day) (kWh)

  • 7.8

0.72 3.84 2852 Des 1 1044 9.62 9.8e+5 2.18 508 Moines 5 304 9.44 4.8e+4 1.89 579 10 218 9.37 1.2e+4 1.71 607 Boston

  • 13.9

1.59 3.79 6034 1 2821 9.91 1.2e+7 2.22 985 5 866 9.91 6.7e+5 2.05 1058 10 527 9.89 1.8e+5 1.88 1111

  • 26.6

4.51 3.81 11944 San 1 4238 10.4 1.7e+8 2.36 1870 Francisco 5 940 10.4 1.6e+6 2.06 2120 10 679 10.7 1.1e+6 2.01 2310

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-26
SLIDE 26

Conclusions and Future Work

  • Exciting era for multi-agent systems in smart homes!
  • Smart Home Device Scheduling Problem and cast it as a DCOP.
  • SH-MGM: a local search-based algorithm to solve SHDS problems.
  • Results:
  • SH-MGM finds better solutions than a simple uncoordinated method.
  • Feasibility established for using a local search-based approach implemented on

hardware with limited storage and processing power.

  • Dataset available for download:

https://github.com/nandofioretto/SHDS_dataset

25

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions

slide-27
SLIDE 27

Conclusions and Future Work

  • Exciting era for multi-agent systems in smart homes!
  • Smart Home Device Scheduling Problem and cast it as a DCOP.
  • SH-MGM: a local search-based algorithm to solve SHDS problems.
  • Results:
  • SH-MGM finds better solutions than a simple uncoordinated method.
  • Feasibility established for using a local search-based approach implemented on

hardware with limited storage and processing power.

  • Dataset available for download:

https://github.com/nandofioretto/SHDS_dataset

26

Thank You!

Smart Homes and SHDS | DCOP | Solution Approach | Results | Conclusions