Skip to main content

Table 4 Energy complexity of each compression algorithm (sum of microprocessor operation and memory accesses)

From: An energy-efficient low-memory image compression system for multimedia IoT products

Type

SPIHT [18]

L-BCWT [23]

ZM-SPECK [25]

VLC_NUQ [26]

FALC [27]

Proposed method

8L JPEG [40]

4L H.264 [13]

4L HEVC [15]

#CPU operations (energy cost/frame)

25 × 3MN (14.7)

15 × 3MN (8.8)

13.9 × 3MN (8.1)

38 × 3MN (22.4)

16 × 3MN (9.4)

58 × 3MN (34.2)

277.6 × 3MN (163.7)

548 × 3MN (323)

653 × 3MN (385)

#Memory accesses (energy cost/frame)

15.5 × 3MN (10.2)

9.5 × 3MN (6.2)

12 × 3MN (7.9)

9 × 3MN (5.9)

6.5 × 3MN (4.2)

43.6 × 3MN (28.8)

37.4 × 3MN (24.7)

194 × 3MN (128)

334 × 3MN (221)

Average FCR (RF energy cost/frame)

2.04:1 (469)

2.04:1 (469)

2.37:1 (404)

3.25:1 (295)

3.37:1 (284)

4.9:1 (195)

11.47:1 (83.4)

10.49:1 (91)

12.5:1 (76.5)

Transmission duration (s)

0.401

0.401

0.345

0.252

0.243

0.167

0.071

0.078

0.065

Total energy cost (mJ/frame)

494.2

484.4

420

322.9

297.8

258.4

271.9

542.6

682.4

  1. M image width size, N image height size. CPU operation energy cost, 62.5 pJ/bit; memory access energy cost, 70 pJ/bit; average RF power consumption, 1170 mW; and RF bandwidth, 11 Mbps