Energy efficient machine to machine communications

KTH ROYAL INSTITUTE OF TECHNOLOGY

Energy Efficient MAC for Cellular-Based M2M Communications Amin Azari and Guowang Miao KTH Royal Institute of Technology

GlobalSIP Conference, 2014, Atlanta, USA

Contents:

• Introduction • System model and problem formulation • Proposed MAC design • Simulation Results • Conclusion

9/8/2015 2

Motivation Future wireless access (5G) • Key challenges

• Continued traffic growth in terms of volume • Continued traffic growth in terms of number of devices • Energy efficient system design

9/8/2015 3

M2M communication

• M2M communications: Communication of smart devices without human intervention.

• Some characteristics: • Large number of short-lived sessions • (usually) low-payload • Vastly diverse characteristics (e.g. battery capacity) • Vastly diverse QoS requirements (e.g. delay)

9/8/2015 4

M2M Communication Enablers

Reliability Av

aila

bilit

y

Cellular-based M2M

Proprietary Cellular

Low-power WLAN

Zigbee-like

Low-power Bluetooth

• Reliability = resilience to interference, throughput and outage guarantees

Reference: GREEN NETWORK TECHNOLOGIES FOR MTC IN 5G, Jesus Alonso-Zarate, EIT/ICT Summer school presentation

• Availability = coverage, roaming, mobility

9/8/2015 5

Coverage Mobility & Roaming Interference Control Energy Efficiency ?

☑ ☑ ☑

Contents: • Introduction • System model and problem formulation • Proposed MAC design • Simulation Results • Conclusion

9/8/2015 6

System model

• Single Cell • N machine nodes

• Battery-driven nodes • Long battery-life is desired

• Specific resource allocation for M2M (no cellular user) • Event-driven and data (Poisson packet arrival)

9/8/2015 7

Problem formulation

• Clustering design • Complete, partial or no-clustering? • Number of clusters • Cluster-head selection & reselection

• Communication Protocol • Intra-cluster communication protocol • Inter-cluster communication protocol

9/8/2015 8

Problem formulation

• Clustering design • Presented in

Energy-Efficient Clustering Design for M2M Communications, G. Miao and P. Zhang, GlobalSIP 2014

• Communication protocol • is discussed in this work

9/8/2015 9

Contents: • Introduction • System model and problem formulation • Proposed MAC design

• Clustering for cellular-based M2M • Intra-cluster communication • Inter-cluster communication

• Simulation Results • Conclusion

9/8/2015 10

Proposed MAC design: Clustering

• Clustering • Given desired receive SNR • Calculate transmission power at ith node, 𝑃𝑖

• If 𝑃𝑖 > 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡

– node i is to be clustered

• In each cluster the node with lowest 𝑃𝑖 will be CH.

9/8/2015 11

Proposed MAC design: Intra-cluster Communication

• Intra-cluster communication • Relatively low-load regime

• CSMA/CA has good performance in low-load regime • Scalable, low signaling overhead, and acceptable EE

• The EE, delay, and user capacity analysis:

9/8/2015 12

Proposed MAC design: Multi-Phase CSMA

• Even more energy efficiency • Multi-phase CSMA for intra-cluster communication • Enables close-to-zero power wastage • Needs local synchronization (tradeoff)

9/8/2015 13

Analytical results regarding EE and delay are presented.

Proposed MAC design: Inter-cluster

• Inter-cluster communication • Heterogeneous system

• Length of data packet (CH and CM) • State: delay critical, queue status and residual energy

• Interference to the cellular users must be avoided. THEN

• Reservation-based protocols (dynamic TDMA) • Moderate scalability and energy-saving

• Analytical results are omitted from the paper due to the page limit.

9/8/2015 14

Proposed MAC: Communication frame

9/8/2015 15

Inter-cluster Intra-cluster

Multi-phase CSMA Reservation phase

Notification phase Transmission phase

Contents: • Introduction • System model and problem formulation • Proposed MAC design • Simulation Results • Conclusion

9/8/2015 16

Simulation Results: System Model

• Single cell with LTE base station • Uplink transmission of 𝑁 battery-driven machine nodes • 4-phase CSMA for intra-cluster communication • Dynamic TDMA for inter-cluster communication • Poisson packet arrival at nodes • Clustering threshold: varied

9/8/2015 17

Simulation Results_1

9/8/2015 18

Partial clustering

Delay and energy performance evaluation

No clustering

Complete clustering

Simulation Results Analysis

9/8/2015 19

• Clustering is not always (for all nodes) EE • However, it always eases the massive access problem

• Partial clustering is optimal • Delay performance is sacrificed for getting EE • Tradeoff delay/energy efficiency

Simulation Results_2

9/8/2015 20

Battery lives of cluster heads (CH) and members (CM) for proposed MAC and dynamic TDMA

Simulation Results Analysis

9/8/2015 21

• Proposed MAC has extended the battery life of nodes.

• The extension is 500% on average and 800% at some points.

• The battery life of cluster heads is sacrificed by 50%.

• Cluster-head reselection scheme

Conclusion

• Key requirement for enabling M2M communication over cellular networks • Providing efficiency

• Energy efficient massive access can prolong the lifetime • Clustering for all nodes is not EE • Using CH reselection algorithms, one can prolong the

overall network lifetime

9/8/2015 22

Future works

• Revisiting design principles of cellular networks to address massive access problem in an efficient way • Considering heterogeneous characteristics of machine

nodes • Considering heterogeneous QoS of machine nodes

9/8/2015 23

Thanks for your participation.

9/8/2015 24

Supporting Materials

9/8/2015 25

Cellular-based M2M M2M communications supported by cellular networks • Direct or through gateway

Advantages: • Ubiquitous Coverage • Mobility & Roaming • Interference Control

Disadvantages: • Designed and optimized for small number of long-lived sessions

• Massive access problem • Energy inefficiency

• Attaching to the network is contention-based, etc. • Physical layer inefficiency

• Not optimized for small data payload

9/8/2015 26

Problem formulation • Access schemes

• Contention-free schemes – Not scalable (High signaling) – High average packet delay – High energy efficiency

• Contention-based schemes – Scalable and distributed – Low-delay in low-load/ High-delay in high-load – Energy wasting in medium- to high-load regime

• Reservation-based schemes – Contention-based in reservation, -free in transmission

9/8/2015 27

Details of the derived performance analyses

9/8/2015 28

𝑔: aggregated traffic arrival rate ps: probability of successful transmission

𝜏𝑠 = 𝜏𝑝+ 𝜏𝑟 𝜏𝑝: packet length 𝜏𝑟: Round trip time from transmission to acknowledgement.

Energy Efficient System Design

• Energy Saving ≠ Energy Efficiency • Complete Saving of Energy = Shut down network

completely to save the most energy • Not desired!

• Purpose of energy-efficient wireless network design • Not to save energy • Make the best/efficient use of energy!

• Energy saving w/o losing service quality • Bit-per-Joule design metric

9/8/2015 29