A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Rafael Perazzo Barbosa Mota IME - USP 05 de junho de 2013
Agenda 1 Motivation and relevance 2 Background 3 Related Work 4 Our Proposal 5 Results and discussion 6 Conclusions 7 References
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Motivation and relevance Table of Contents 1 Motivation and relevance 2 Background 3 Related Work 4 Our Proposal 5 Results and discussion 6 Conclusions 7 References 3 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Motivation and relevance Motivation and relevance Radio Frequency Identification (RFID) is a key technology of IoT since small passive RFID tags make it possible to link millions and billions of physical products with Internet [1]. 2 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Motivation and relevance Motivation and relevance Therefore, RFID tag anticollision algorithms will play an important role in IoT [1]. 3 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Motivation and relevance Background - DFSA 4 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Background Table of Contents 1 Motivation and relevance 2 Background 3 Related Work 4 Our Proposal 5 Results and discussion 6 Conclusions 7 References 5 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Background Dynamic Frame Slotted Aloha - DFSA Algorithm 1 DFSA algorithm Require: L ⊲ L is the initial frame size 1: continue ← true 2: n ← L 3: repeat ⊲ While collisions occurs i ← 0 ⊲ Initial slot time 4: counter ← 0 ⊲ Number of received replies (=1, =0 or > 1) 5: collisions ← 0 ⊲ Collision counter 6: for i ≤ n do ⊲ Sends every slot time 7: Query(n,i) ⊲ Sends a Query Command with frame size n and 8: slot i Wait for reply 9: if ( counter == 1) then 10: QueryRep() ⊲ Reader sends an ACK to identify the tag 11: else if ( counter > 1) then 12: collisions ← colisions + 1 13: end if 14: end for 15: if ( collisions == 0) then 16: continue ← false 17: else 18: n ← Call a function to calculate the next frame size 19: L ← n 20: end if 21: 22: until (continue==true) 6 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Related Work Table of Contents 1 Motivation and relevance 2 Background 3 Related Work 4 Our Proposal 5 Results and discussion 6 Conclusions 7 References 7 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Related Work Q Algorithm [2] 8 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Related Work Schoute [3] Algorithm 2 Schoute algorithm 1: function Schoute( collisions ) return round ( 2 . 39 ∗ collisions ) 2: 3: end function 9 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Related Work Mota 2013 Algorithm 3 Mota algorithm 1: function mota( collisions ) return round ( 2 . 62 ∗ collisions ) 2: 3: end function 10 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Related Work Eom-Leee [4] Algorithm 4 Eom-Lee Estimation Algorithm 1: function estimation_eomlee( ǫ, collisions , success ) ⊲ collisions and success are the number of collision and success slots in last frame, respectively. ǫ is the stop criteria b 1 ← ∞ 2: y 1 ← 2 3: backlog ← L 4: repeat 5: backlog b prox ← 6: y 1 ∗ collisions + success 7: − 1 1 − e bprox y prox ← 8: − 1 bprox ) 1 b prox ∗ ( 1 − ( 1 + bprox ) ∗ e 9: backlog ← y prox ∗ collisions 10: temp ← y 1 11: y 1 ← y prox 12: b 1 ← b prox 13: until ( | y 1 − temp | < ǫ ) 14: 15: return round(backlog) 16: end function 11 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Related Work Others Dynamic, Adaptative and Splitting BTSA (Excellent System Efficiency. Many changes must be done on tags operation) [1] 12 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Related Work Others Dynamic, Adaptative and Splitting BTSA (Excellent System Efficiency. Many changes must be done on tags operation) [1] Vogt (equivalent to Q Algorithm) [5] 12 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Related Work Others Dynamic, Adaptative and Splitting BTSA (Excellent System Efficiency. Many changes must be done on tags operation) [1] Vogt (equivalent to Q Algorithm) [5] Chen (worse than Q Algorithm) [6] 12 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Our Proposal Table of Contents 1 Motivation and relevance 2 Background 3 Related Work 4 Our Proposal 5 Results and discussion 6 Conclusions 7 References 13 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Our Proposal EDFSA-I Estimate initial START frame size L slot=0 Send Query command and current slot >1 0 idle++ Tags replies? collisions++ slot++ slot++ No 1 No End of End of Frame ? Frame ? Send ACK slot++ Yes Yes Calculate new Frame Size using Eom-Lee method slot=0 14 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Our Proposal EDFSA-II 15 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Our Proposal EDFSA-II Algorithm 5 Estimated Dynamic Framed Slotted Aloha - EDFSA Require: ǫ ⊲ ǫ is the stop criteria 1: L ← estimation ( 1 , 3 ) ∗ 0 . 67 2: i ← 1 ⊲ Initial slot time 3: counter ← 0 ⊲ Number of received replies (=1, =0 or > 1) 4: collisions ← 0 ⊲ Collision counter 5: for ( i = 1 ; i ≤ L ; i ← i + 1 ) do ⊲ Sends every slot time Query(L,i) ⊲ Sends a Query Command with frame size n and slot i 6: Wait for reply 7: if ( counter == 1) then 8: QueryRep() ⊲ Reader sends an ACK to identify the tag 9: success ← success + 1 10: else if ( counter > 1) then 11: collisions ← collisions + 1 12: resolve _ collisions () ⊲ Collisions are resolved as soon as they 13: occur end if 14: 15: end for 16 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Our Proposal How to estimate the initial frame [7] ? collisions=0 START idle=0 success=0 Q++ Send i times collisions=i or idle=i or QueryEst command Yes Q-- No Yes FinalQ=Q collisions=i or idle=i Send i times QueryEst No FinalQ is the estimated number of tags END 17 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Our Proposal How to resolve local collisions ? START L=3 Next slot Send Query with slot i Wait for replies >1 =1 QueryRep() Number of Next slot collisions++ replies ? success++ No Yes End of collisions = 0 ? L = mota(collisions) frame ? Yes END 18 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Our Proposal How to select best parameters for estimation ? 15 c=1 and i=3 c=1 and i=5 14 c=0.3 and i=3 c=0.2 and i=3 13 c=0.2 and i=5 12 11 10 Best Initial Q Value 9 8 7 6 5 4 3 2 1 0 200 400 600 800 1000 1200 1400 1600 1800 Number of tags 19 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Our Proposal How to select best parameters for estimation ? 400 c=1 and i=3 c=1 and i=5 375 c=0.3 and i=3 350 c=0.2 and i=3 c=0.2 and i=5 325 300 275 250 Delay (slots) 225 200 175 150 125 100 75 50 25 0 50 250 450 650 850 1050 1250 1450 1650 1850 Number of tags 20 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Our Proposal How to select the best initial frame size for resolve local collisions ? 100 % 95 % 90 % 85 % 80 % 75 % Mean: aprox.(2.616) 2.62 (2.633) 70 % Sample size: 88751 collisions slots 65 % Confidence Interval (CI): 99% 60.2082 60 % Frequency 55 % 50 % 45 % 40 % 35 % 30 % 25.0518 25 % 20 % 15 % 9.5896 10 % 5 % 3.3430 1.1797 0.4124 0.1532 0.0338 0.0169 0.0090 0.0023 0 % 2 3 4 5 6 7 8 9 10 11 12 Number of tags in collision 21 / 31
A New Dynamic Frame Slotted Aloha Anti-Collision Algorithm for the Internet of Things Our Proposal How to select the next frame estimation method ? 460 % Lower Bound 440 % Schoute 420 % 429 Eom−Lee 400 % Mota 380 % Average Difference compared to Q Algorithm 360 % 371 PS. Negative number means a decrease on System Efficiency 357 340 % PS. Positive number means a increase on System Efficiency 336 320 % Initial Frame Size: 3 300 % 280 % 260 % 240 % 242 220 % 200 % 211 200 195 180 % 160 % 140 % 120 % 122 117 113 113 100 % 80 % 84 80 60 % 72 72 57 40 % 46 39 39 37 33 20 % 30 27 27 27 23 0 % −3 −20 % 2 3 4 5 6 7 8 Number of tags in collision 22 / 31
Recommend
More recommend
