Transcript Channel-Independent Viterbi Algorithm (CIVA) for DNA

Make Random Access Contentions
Transparent by Orthogonal
Complementary Codes in Wireless
Communications
Xiaohua (Edward) Li
Department of Electrical and Computer Engineering
State University of New York at Binghamton
Outline
•
•
•
•
•
Introduction
Access request and detection
Performance analysis
Simulations
Conclusions
Introduction: Random Access
• Random access: efficient for heterogeneous traffic
• Problems of random access: contention
– Throughput loss  severe in high traffic load
– Delay increased  difficult to maintain QoS
• Resolve contention to improve efficiency
Introduction: Contention Resolution
• Some traditional methods
– Slotted ALOHA, CSMA/CA
– Reservation-ALOHA, RTS/CTS
– TDMA-CDMA
• Common characteristics
– Treat problem in MAC layer only
– Collided packets simply discarded
– Do not utilize information of physical layer signals
Introduction: Contention Resolution
• Alternative methods: separate collided signal
– Physical layer signal processing
– By, e.g., repeated transmission, multi-user detection,
constant modulus, etc
• Difficulties
– Signal separation is difficult, suffers many practical
problems, e.g., ill-channel conditions, complexity
• Solution:
– joint physical/MAC layer design
Proposed Method: Basic Idea
• Use access request packets (ARP)
• Physical layer:
– Separate collided (ARP) only
– Make collision transparent to MAC
• MAC layer:
– Schedule transmission of ARP and data packets
– Make physical layer signal separation easy
• Orthogonal complementary codes:
Efficient and robust collision separation
System Design
• Slotted channel: access request slot, data slot
• All active users transmit ARP in the same
access request slot
• Contentions exist in access request slot only,
not in data slot
Orthogonal Complementary Code
• OC code set properties:
–
–
–
–
with: I flocks, J family/flock, L-bit code/family
Processing gain: JL
Orthogonal among flocks, irrespectively shifting
Orthogonal within each flock with non-zero shifting
n
0

1 
  ckj (0) 
J 1




 cij (0)  cij ( L  1)         JL i  k  n
j 0

 ckj ( L  1)
1 0

n  0,1,2,; 0  i  I  1, 0  k  I  1


Access Request Packets
• Designed with OC codes
bTi ,d  [0dG ci ,0 0DG  ci , J 2 0DG ci , J 1 0( Dd 1)G ]
• Packet (slot) length: Lp  (U  1)G  JLc
• Efficient for large number of users
Access Request Detection
 Detectorfor the user um,k :
 0l 
1 

Fm,k  [f m,k (0)  f m,k (G  1)], f m,k (l ) 
b
m, k 
JLc 
0G  l 1 
 Decisionmetric :

H
zm,k ( n)  y ( n)Fm,k


Joint Physical/MAC Layer Design
• Protocol
– At the beginning of a frame, central controller asks
for access request
– Active users transmit ARP
– Central controller detects access requests
– Assign data packet slots to active users
• Properties:
– Efficient ARP structure, with user ID inherently
embedded
– ARP collision separation: efficient and robust to
asynchronous, near-far, multipath
Performance Analysis
 data/access - requestslotlengthLd / La , Traffic load 
 Markovchain model : number of data slot/frame  state
p j  P ( i | j )   P ( j | i ) pi ,
i j
j  0,1,
i j
 Throughtand delay (without ARP detectionerror) :
R



j 0

(
j 0
jLd p j
jLd  La ) p j
, T

j ( j  1) j ( j  1) 


2
 2

 j 0 jp j
 
j 0 
Consider Detection Error
 Two typesof detectionerrors : P 1 , P 2
 Transitional probabilities :
P1 (i | j )  P (
i  jP 1
| j ),
1  P 1
P3 (i | j )  P (
(i  jP 1 )(1  P 2 )
| j)
1  P 1
P2 (i | j )  P (i  jP 2 | j ),
 Throughputs :
R1 



j 0

(
j 0
jLd p1 j
jLd  La ) p1 j
, R2 

j (1  P 2 ) Ld p3 j
j 0

( jLd  La ) p3 j
j 0

R2 


j (1  P 2 ) Ld p2 j
j 0

( jLd  La ) p2 j
j 0


Consider Detection Error
 Delays:
T1 
T2 
T3 

 
j 0 
j ( j  1) j ( j  1) j ( j  1) P 1 


 p1 j
2
1  P 1 
 2
 2,

 j 0 jp1 j


j ( j  1) j ( j  1) 

p2 j

2
 2

 2,

 j 0 j(1  P 2 ) p2 j
 
j 0 
 
j 0 
j ( j  1)(1  P 2 ) j ( j  1) j ( j  1) P 1 (1  P 2 ) 


 p3 j
2
2
1  P 1


 2,

 j 0 j(1  P 2 ) p3 j
Simulations: Throughput
Compare throughput: theory and
simulated
Simulations: Delay
Compare delay: theory and
simulated
Simulations: ARP Detection
• 60 users, guard
length 10,
• Processing Gain 64,
random channel
with max length 5
• Random
asynchronous delay
(max 5)
• Random near-far
(NF)
Simulations: ARP Detection
• Decision error rate and
traffic load
Conclusions
• Joint physical/MAC layer design to
– Resolve contentions, to improve efficiency
– Make contentions transparent, to support QoS
• Access request collision resolution with OC
codes
– Efficient in computational complexity
– Robust to (ill) multipath channels, near-far &
asynchronous transmission