A Multiagent System Approach to Schedule Devices in Smart Homes William Yeoh Enrico Pontelli Ferdinando Fioretto New University State University University of Michigan Feb, 2017
Home Automation Fig.1 Fig.2 1
Network of smart homes Fig.3 2
SHDS Results Conclusions Background Outline • Background (DCOPs) • Smart Homes Device Scheduling (SHDS) • Results • Conclusions and Future work 3
SHDS Results Conclusions Background Distributed Constraint Optimization < 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. x a x a x b cost 0 0 3 f ab f ac 0 1 ∞ 1 0 2 1 1 5 x b x c f bc Constraint graph Constraint (cost table) 4
SHDS Results Conclusions Background Distributed Constraint Optimization < 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 min F ( x ) x min X = arg max f ( x | scope ( f ) ) x F f 2 F 5
SHDS Results Conclusions Background DCOP: Assumptions • Agents coordinate an assignment for their variables. x a • Agents operate distributedly. f ac f ab Communication: • By exchanging messages. x b x c f bc • Restricted to agent’s local neighbors. Knowledge: f bd • Restricted to agent’s sub-problem. x d • Privacy preserving. 6
SHDS Results Conclusions Background Smart Home Device Scheduling (SHDS) A SHDS problem is composed of: H : A neighborhood of smart homes. • Z i : A set of smart electric devices within each home h i . • H: A time horizon for the device scheduling. • exact 2 Z i . The ener problem uildings h i and whose 7
SHDS Results Conclusions Background Smart Home Device Scheduling (SHDS) A SHDS problem is composed of: H : A neighborhood of smart homes. • Z i : A set of smart electric devices within each home h i . • H: A time horizon for the device scheduling. • θ : A pricing function expressing cost per kWh of energy consumed. • 8
SHDS Results Conclusions Background Smart Home A smart home h i has: A set of smart devices Z i it can control. • exact 2 Z i . The ener problem uildings h i and whose 9
SHDS Results Conclusions Background Smart Home A smart home h i has: A set of smart devices Z i it can control. • L i A set of locations (e.g., living room, kitchen) • 10
SHDS Results Conclusions Background Smart Home A smart home h i has: A set of smart devices Z i it can control. • L i A set of locations (e.g., living room, kitchen) • P H A set of state properties (e.g., cleanliness, temperature) • Battery charge sensor Thermostat Cleanliness sensor 11
SHDS Results Conclusions Background Smart Devices (Actuators) A smart device is defined with a Location, defining the place where the device can act (e.g., living room) • The possible actions it can perform (clean, charge, stop) and the power • consumption associated to them The set of state properties it affects (e.g., cleanliness, battery_charge) • Action State property Power (kW/h) run cleanliness 0.0 charge battery charge 0.26 stop 0.0 12
SHDS Results Conclusions Background 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. Thermostat Heater Oven Current Next State State off off 12 C 11 C Effect of the environment off bake 12 C 13.8 C on off 12 C 17.5 C on bake 12 C 19.3 C 13
SHDS Results Conclusions Background Smart Device Schedules Scheduling Rules Simple syntax to express scheduling rules: • h location i h state property i h relation i h state i h time i Active rules : specify user-defined objectives on a desired state • of the home. E.g., living room cleanliness ≥ 75 before 1800 Passive rules : define implicit constraints on devices. E.g., • z v battery charge ≥ 0 always z v battery charge ≤ 100 always 14
SHDS Results Conclusions Background Smart Device Schedules Schedule: A sequence of actions for the home devices. S R R C C R R C R Device Schedule 0 15 30 30 30 45 60 60 75 Cleanliness (%) 65 40 15 35 55 30 5 25 0 Battery Charge (%) Goal 75 75 Battery Charge (%) 60 60 Cleanliness (%) 45 45 30 30 15 15 0 0 1400 1500 1600 1700 1800 Time Deadline 15
SHDS Results Conclusions Background Smart Home Device Scheduling (SBDS) • SHDS objective: Aggregated monetary cost of the homes schedules ↵ c · C sum + ↵ e · E diff min ξ [0 ! H ] Z i Energy consumption peaks across all homes Homes’ devices schedules subject to: ⇠ [ t a ! t b ] 8 h i 2 H , R [ t a ! t b ] = R [ t a ! t b ] 2 R i : | p p Φ p P All scheduling rules must be satisfied 16
SHDS Results Conclusions Background DCOP mapping SBDS DCOP • A home h i ϵ H. • Agent a i ϵ A • Variable x i ϵ X (controlled by a i ) • A device z j (in building h i ) • Action j for device z j. • j-th value in domain D i of variable x i • Schedule costs for a device z j • Local soft constraint • Device scheduling feasibility • Local hard constraint • Energy peak consumption • Global soft constraint 17
SHDS Results Conclusions Background Solution Approach SH-MGM: Adaptation of a local search DCOP algorithm (MGM). 1. Agents independently search for a feasible schedule for their local devices. exact 2 Z i . The ener problem uildings h i and whose c i : schedule cost E it :energy consumption 18
SHDS Results Conclusions Background 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. exact (c i , E it ) 2 Z i . The ener problem uildings h i and whose (c i , E it ) c i : schedule cost E it :energy consumption 19
SHDS Results Conclusions Background Solution Approach SH-MGM: Adaptation of a local search DCOP algorithm (MGM). 3. Upon receiving all other agents costs and energy consumptions: exact (c j , E jt ) 2 Z i . The ener problem uildings h i and whose (c k , E kt ) c i : schedule cost E it :energy consumption 20
SHDS Results Conclusions Background 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. lem with its current solution, exact α c · c i ( ξ [0 ! H ] ) + α e · E diff 2 Z i . Z i The ener Then, within a given time limit, current schedule problem uildings h i and whose c i : schedule cost E it :energy consumption 21
SHDS Results Conclusions Background 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. new schedule E diff ≤ α c · c i ( ξ [0 ! H ] ξ [0 ! H ] α c · c i (ˆ ) + α e · ˆ ) + α e · E diff Z i Z i current schedule c i : schedule cost E it :energy consumption 22
SHDS Results Conclusions Background 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 G i between its current and new solutions, and broadcast it to all other agents. ⇣ ) + α e · E diff ⌘ α c · c i ( ξ [0 ! H ] G i = Z i ⇣ E diff ⌘ ξ [0 ! H ] α c · c i (ˆ ) + α e · ˆ − Z i c i : schedule cost E it :energy consumption 23
SHDS Results Conclusions Background Solution Approach SH-MGM: Adaptation of a local search DCOP algorithm (MGM). 4. Upon receiving all other agents’ gains G k , 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. exact 2 Z i . G j The ener problem uildings h i and whose G k G i : agent’s gain 24
SHDS Results Conclusions Background 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. A Raspberry PI with a dangle Smart devices 25
Recommend
More recommend