Lifetime-Aware Scheduling and Power Control for Cellular-based M2M Communications

KTH ROYAL INSTITUTEOF TECHNOLOGY

Lifetime-Aware Scheduling and Power Control for Cellular-based M2M Communications

Amin Azari and Guowang Miao

KTH Royal Institute of Technology

WCNC Conference 2015, USA

04/15/2023 2

Contents:

• Introduction• System model and problem formulation• Proposed Scheduling and Power Control• Low-complexity uplink scheduling• Performance evaluation• Conclusion

04/15/2023 3

IntroductionFuture wireless access (5G) • Key challenges:

• Continued traffic growth in terms of volume,• Continued traffic growth in terms of

number of devices,• Spectral & Energy efficient system design.

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)

04/15/2023 4

System model: Basics

• Single Cell with one BS in the center• Machine nodes

• Semi constant data-rate generation• Battery-driven nodes, long battery-life is desired

• SC-FDMA for uplink transmission to the BS [1]:• Each subcarrier can only be allocated to at most one user.• Only adjacent-subcarriers can be allocated to a user.• The transmit power on all subcarriers assigned to a user must be the

same.• In our previous works, optimal cluster-forming, cluster-head selection,

and communication protocol design for cellular-based M2M is investigated to settle the massive access problem [2]-[3].

• Here we investigate the scheduling for the CHs and unclustered nodes to further increase the network lifetime.

04/15/2023 5

System model: Lifetime Metric Definition:

Constant energy consumption for synchronization, admission control, and data gathering in each duty cycle

Energy consumption in sleep mode in each duty cycle

Energy consumption in each duty cycle for reliable data transmission with non-perfect power amp + circuit energy consumption)

Duty cycle of operation for node

Expected lifetime for node at time

Remaining energy of node at time

𝐸𝑡=𝐷 𝑖

𝑅𝑖

(𝑃𝑐+𝛼 𝑃 𝑡𝑖)

04/15/2023 6

System model: Network Lifetime Definition

We consider three definitions for the network lifetime:

1. The average length of individual lifetimes,

2. The shortest length of individual lifetimes,• This is applicable when missing even one node

deteriorates the performance or coverage.

3. The longest length of individual lifetimes,• This is applicable when the correlation between

gathered data by the sensors is high.

4. The solution to the (3) follows the same procedure as (2).

04/15/2023 7

Problem formulation (Scheduling and

power control)

• Finding• optimal number of assigned chunks to user • : optimal transmit power for user

• where, Packet length; Data rate Total transmission time Number of subcarriers : Total number of available chunks

Optimization problem:

04/15/2023 8

Heuristic solution

• We have two optimization problems for maximizing network lifetime under two lifetime definitions.

• NP-hard mixed integer scheduling• Joint scheduling and power control is non-convex.

• We present two heuristic solutions:• Algorithm 1: Average individual lifetime maximization.• Algorithm 2: Minimum individual lifetime maximization.

• The procedure for both algorithms is as follows:

04/15/2023 9

Heuristic solution: Brief description of Algorithm 1 and 2.

1. Find the set of possible chunk allocation s.t. the constraints.

2. Calculate the optimal transmit power under each chunk allocation.

• The individual lifetime is a quasiconcave function of and we derived the optimal transmit power for node i on each subcarrier as:

04/15/2023 10

Heuristic solution: Brief description of Algorithm 1 and 2.

1. Find the set of possible chunk allocation s.t. the constraints.

2. Calculate the optimal transmit power under each chunk allocation.

3. Find the set of valid chunk allocations by comparing their derived transmit power against the constraints on the transmit power.

4. Calculate the lifetime for each device and each chunk allocation

5. Find the chunk allocation which maximizes the network lifetime.

6. Return the chunk allocation and its respective transmit power.

04/15/2023 11

Heuristic solution: Complexity Analysis

• Complexity order• Algorithm 1:• Algorithm 2:

• There are M2M scenarios that this complexity is meaningful.• Example:

– the time-controlled scheduling for enabling M2M communications in cellular networks in which each class of the nodes is assigned a constant amount of resources in regular intervals.

04/15/2023 12

Heuristic solution: Complexity Analysis

• Using algorithms 1 and 2, the designed scheduling by the BS for each class of nodes will be valid for a long time-interval

• This is because: • The number and position of machine devices are semi-

constant due to lack of mobility of most machine devices;

• The energy consumption of machine devices is expected to be low, then the change in remaining energy will be low;

• The packet length of machine nodes is constant in most M2M applications.

• Then, finding the optimal solutions which are valid for a long time-interval, even with high complexity, will be meaningful.

04/15/2023 13

Low-complexity Solutions

• The transmit energy-per-subcarrier is assumed to be the same for all nodes.• The transmit power is controlled by the number of assigned

subcarriers/chunks to each node.• The lifetime expression is a concave function of number of

assigned chunks.

• Using linear relaxation, we transform the NP-hard integer scheduling problem into a related linear convex optimization problem.• Solvable in polynomial time,• Much lower complexity than presented algorithms.

04/15/2023 14

Low-complexity Solutions

• Using dual Lagrangian method:• Optimal scheduling when the average individual lifetime is

defined as the network lifetime:

• Optimal scheduling when the minimum individual lifetime is defined as the network lifetime:

where:

04/15/2023 15

Simulation Results• We adopt the time-controlled frame work for enabling M2M

communication in cellular networks.

• 0.6 MHz bandwidth is allocated to a class of M2M machines (with the same duty cycle) with 10 members in each duty cycle.

• The M2M network and scheduling protocols are implemented in the MATLAB software.

• Benchmarks:• Equal resource allocation (ERA)• Throughput-aware RA, in which closer nodes have more

chance to receive more resource blocks.

04/15/2023 16

200 300 400 500 600 700 800 900 1000

1

1.2

1.4

1.6

1.8

Packet Length (Di+D

oh) in bits

Life

time

Fa

cto

r

200 300 400 500 600 700 800 900 1000

1.02

1.04

1.06

1.08

1.1

x 107

Life

time

(

duty

cyc

le)

Optimal LT-awareSubopt. LT-awareEqual RAThroughput-awareEqual RA (abs. value)

Simulation Results

• Lifetime factor for scheme x: absolute lifetime under scheme x divided by the absolute lifetime under ERA.

• Network lifetime: Min individual lifetime Algorithm 2 is used.

Absolute Lifetime under ERA

Lifetime factor

04/15/2023 17

Simulation Results Analysis

• The heuristic solution from algorithm 2 outperforms the others.• The performance of low-complexity solution is close to the

optimal one.• The lifetime-aware solutions outperforms the other schemes

significantly.

04/15/2023 18

Conclusion

• Key requirement for enabling M2M communication over cellular networks:• Providing energy efficiency,• Energy efficiency maximization individual lifetime

maximization.• Energy efficient scheduling and power control can prolong the

network lifetime significantly.

04/15/2023 19

References:

• H. Myung, J. Lim, and D. Goodman, “Single carrier FDMA for uplink wireless transmission,” IEEE Vehicular Technology Magazine,, vol. 1, no. 3, pp. 30–38, Sept. 2006.

• A. Azari and G. Miao, “Energy efficient MAC for cellular-based M2M communications,” in 2nd IEEE Global Conference on Signal and Information Processing, 2014.

• P. Zhang and G. Miao, “Energy-Efficient Clustering Design for M2M Communications,” in 2nd IEEE Global Conference on Signal and Information Processing, 2014.

04/15/2023 20

Thanks for your participation.

Question and Answer