backscatter-aided hybrid data offloading for wireless ...19-hybrid-relay.pdf · y. zou et al....

Backscatter-Aided Hybrid Data Offloading for

Wireless Powered Edge Sensor Networks

Yuze Zou, PhD

School of Electronic Information and Communications

Huazhong University of Science and Technology

IEEE Global Communications Conference

9-13 December 2019

Waikoloa, HI, USA

Revolutionizing Communications

Y. Zou et al. Backscatter-Aided Hybrid Data Offloading for Wireless Powered Edge Sensor Networks

Introduction

System Model

Energy Minimization for Hybrid Data

Offloading

Numerical Results

2/18

Outline


Introduction

System Model


Offloading

Numerical Results

3/18

Outline


Introduction

4/18Img src: https://www.newgenapps.com/blog/influence-of-iot-on-wearable-technology

MECServer

Cloud Server

Edge Devices


Communication models in the edge sensor networks

Active Transmission

Relative high transmit rate

High energy consumption

Passive Transmission

Relative low transmit rate

Negligible energy consumption

Introduction

5/18


Introduction

System Model


Offloading

Numerical Results

6/18

Outline


System Model

7/18

𝑆𝑖

𝑆𝑗

⋯

⋯⋯

𝑆𝑁

𝑆1

𝑔𝑖 ℎ𝑁

HAP


Time Allocation

Each sensor is allocated with fixed slot, say 𝑇/𝑁

Each sensor operate in passive and then active mode

Each sensor harvests energy when others in passive mode

8/18

System Model


Signal Model Active communications

𝑟𝑎,𝑖 = 𝐵 log2 1 +𝑝𝑎,𝑖 ℎ𝑖

2

𝜎2

Passive communication

𝑟𝑝,𝑖 = 𝑟𝑝

NOTE: 𝑟𝑎,𝑖 > 𝑟𝑝,𝑖 holds true in general.

9/18

System Model

Constant

𝑝𝑎,𝑖 = 𝛽 𝑟𝑎,𝑖≜ 2𝑟𝑎,𝑖/𝐵 − 1 𝜎2/ ℎ𝑖

2⟺ Power

Consumption


Introduction

System Model


Offloading

Numerical Results

10/18

Outline


11/18

Energy Minimization for Hybrid Data Offloading

min𝒕𝒂,𝒕𝒑

𝒊∈𝒩

𝑝0𝑡𝑝,𝑖

s. t. 𝑡𝑎,𝑖 + 𝑡𝑝,𝑖 ≤ 𝑇/𝑁

𝑙𝑎,𝑖 + 𝑙𝑝,𝑖 ≥ 𝐿𝑖

𝑡𝑎,𝑖 ෨𝛽𝑙𝑎,𝑖𝑡𝑎,𝑖

≤ 𝜂

𝑗∈𝒩𝑖

𝑝0𝑔𝑖2𝑡𝑝,𝑗

𝑡𝑎,𝑖 ≥ 0, 𝑡𝑝,𝑖 ≥ 0, 𝑙𝑎,𝑖 ≥ 0

Slot limitation

Task fulfillment

Energy constraint

Energy minimization at HAP

Physical constraint

෨𝛽 𝑟𝑎,𝑖 = 𝛽 𝑟𝑎,𝑖 + 𝑝𝑐,𝑖 denotes the energy consumption of sensor in active mode

𝜂: energy harvesting coefficient, 𝒩𝑖 = 𝒩\ 𝑖𝐿𝑖: total bits to offload of sensor 𝑖, 𝑙𝑎,𝑖 and 𝑙𝑏,𝑖: bits offloaded via active and passive mode, respectively



12/18

Distributed reformulationmin𝒕𝒂,𝒕𝒑

𝒊∈𝒩

𝑝0𝑡𝑝,𝑖

s. t. 𝑡𝑎,𝑖 + 𝑡𝑝,𝑖 ≤ 𝑇/𝑁

𝑙𝑎,𝑖 + 𝑙𝑝,𝑖 ≥ 𝐿𝑖

𝑡𝑎,𝑖 ෨𝛽𝑙𝑎,𝑖𝑡𝑎,𝑖

≤ 𝜂

𝑗∈𝒩𝑖

𝑝0𝑔𝑖2𝑡𝑝,𝑗

𝑡𝑎,𝑖 ≥ 0, 𝑡𝑝,𝑖 ≥ 0, 𝑙𝑎,𝑖 ≥ 0

𝐸𝑖

Sensor 𝑖 is

aware of its

energy supply,

denote by 𝐸𝑖

> This motivates us to optimize sensors’ offloading scheme

in a distributed manner.



13/18

min𝑡𝑎,𝑡𝑝,𝑟𝑎

𝑡𝑝

s. t. C1 𝑡𝑎 + 𝑡𝑝 ≤ 𝑡

C2 𝑡𝑎𝑟𝑎 + 𝑡𝑝𝑟𝑝 ≥ 𝐿

C3 𝑡𝑎𝛽 𝑟𝑎 + 𝑝𝑐𝑡𝑎 ≤ 𝐸

Closed-form solution can be

obtained for each sensor, thus

achieves 𝒪 1 complexity


Introduction

System Model


Offloading

Numerical Results

14/18

Outline


Numerical Results

15/18

HAP’s transmit power: 𝑝𝑡 = 100 mW

Frame length: 𝑇 = 10

Sensor workload: 𝐿 = 10 kbits

Energy harvesting efficiency: 𝜂 = 0.8

Circuit power: 𝑝𝑐 = 1 μW

Passive data rate: 𝑟𝑝 = 5 kbps

Downlink channel gain: 𝑔 2 = −53 dB

Uplink channel gain: ℎ 2 = −60 dB

Noise power: 𝜎2 = −70 dBm

Bandwidth: 𝐵 = 400 kHz

# sensors: 𝑁 = 10

Parameter Settings


Numerical Results

16/18

The HAP’s energy consumption and

transmit time allocation

Can not offload the

workload in time


Numerical Results

17/18

Workload and

transmit time

of passive

data

offloading


Questions & Answers

Thank you !

Q&A

18/18

backscatter-aided hybrid data offloading for wireless ...19-hybrid-relay.pdf · y. zou et al....

Documents