Division algebras Coding for wireless relay networks Conclusion
Division algebras for coding in multiple antenna
channels and wireless networks
Fr´ed´erique Oggier
California Institute of Technology
Cornell University, School of Electrical and Computer Engineering, March 13rd 2007
Division algebras Coding for wireless relay networks Conclusion
Outline
Division algebras
Motivation: coherent space-time coding Introducing division algebras
Other space-time coding applications Coding for wireless relay networks
Distributed space-time coding
Noncoherent distributed space-time coding Conclusion
Division algebras Coding for wireless relay networks Conclusion
The multiple antenna channel (I)
1. Time t = 1:
• 1st receive antenna:y11=h11x11+h12x21+v11 • 2nd receive antenna:y21=h21x11+h22x21+v21
2. Time t = 2:
• 1st receive antenna:y12=h11x12+h12x22+v12 • 2nd receive antenna:y22=h21x12+h22x22+v22
Division algebras Coding for wireless relay networks Conclusion
The multiple antenna channel (I)
1. Time t = 1:
• 1st receive antenna:y11=h11x11+h12x21+v11 • 2nd receive antenna:y21=h21x11+h22x21+v21 2. Time t = 2:
• 1st receive antenna:y12=h11x12+h12x22+v12 • 2nd receive antenna:y22=h21x12+h22x22+v22
Division algebras Coding for wireless relay networks Conclusion
The multiple antenna channel (II)
We get the matrix equation
y11 y12 y21 y22 = h11 h12 h21 h22 x11 x12 x21 x22 | {z } space-timecodeword X + v11 v12 v21 v22 .
Division algebras Coding for wireless relay networks Conclusion
Code design criteria (Coherent case)
• Reliabilityis modeled by the pairwise probability of error:
P(X→Xˆ)≤ const |det(X−Xˆ)|2M.
• We assume the receiver knows the channel (coherent case).
• We need fully diversecodes, such that
det(X−X0)6= 0 ∀ X6=X0. • Diversity reflects the slope of the probability of error. • We attempt to maximize theminimum determinant
min
X6=X0|det(X−X
0
Division algebras Coding for wireless relay networks Conclusion
Code design criteria (Coherent case)
• Reliabilityis modeled by the pairwise probability of error:
P(X→Xˆ)≤ const |det(X−Xˆ)|2M.
• We assume the receiver knows the channel (coherent case). • We need fully diversecodes, such that
det(X−X0)6= 0 ∀ X6=X0.
• Diversity reflects the slope of the probability of error. • We attempt to maximize theminimum determinant
min
X6=X0|det(X−X
0
Division algebras Coding for wireless relay networks Conclusion
Code design criteria (Coherent case)
• Reliabilityis modeled by the pairwise probability of error:
P(X→Xˆ)≤ const |det(X−Xˆ)|2M.
• We assume the receiver knows the channel (coherent case). • We need fully diversecodes, such that
det(X−X0)6= 0 ∀ X6=X0. • Diversity reflects the slope of the probability of error.
• We attempt to maximize theminimum determinant min
X6=X0|det(X−X
0
Division algebras Coding for wireless relay networks Conclusion
Code design criteria (Coherent case)
• Reliabilityis modeled by the pairwise probability of error:
P(X→Xˆ)≤ const |det(X−Xˆ)|2M.
• We assume the receiver knows the channel (coherent case). • We need fully diversecodes, such that
det(X−X0)6= 0 ∀ X6=X0. • Diversity reflects the slope of the probability of error. • We attempt to maximize theminimum determinant
min
X6=X0|det(X−X
0
Division algebras Coding for wireless relay networks Conclusion
Previous work
1. E. Telatar,Capacity of multi-antenna Gaussian channels, 1999.
2. V. Tarokh and N. Seshadri and A. R. Calderbank,Space-time codes for high data rate wireless communications:
Performance criterion and code construction, 1998.
3. B. Hassibi and B.M. Hochwald, High-Rate Codes That Are Linear in Space and Time, 2002.
4. H. El Gamal and M.O. Damen, Universal space-time coding, 2003.
Division algebras Coding for wireless relay networks Conclusion
The idea behind division algebras
• The difficulty in buildingC such that
det(Xi−Xj)= 0,6 Xi 6=Xj ∈ C,
comes from the non-linearityof the determinant.
• IfC is taken inside an algebraof matrices, the problem simplifies to
det(X)6= 0, 06=X∈ C. • A division algebrais a non-commutative field.
Division algebras Coding for wireless relay networks Conclusion
The idea behind division algebras
• The difficulty in buildingC such that
det(Xi−Xj)= 0,6 Xi 6=Xj ∈ C,
comes from the non-linearityof the determinant. • IfC is taken inside an algebraof matrices, the problem
simplifies to
det(X)6= 0, 06=X∈ C.
Division algebras Coding for wireless relay networks Conclusion
The idea behind division algebras
• The difficulty in buildingC such that
det(Xi−Xj)= 0,6 Xi 6=Xj ∈ C,
comes from the non-linearityof the determinant. • IfC is taken inside an algebraof matrices, the problem
simplifies to
det(X)6= 0, 06=X∈ C. • A division algebrais a non-commutative field.
Division algebras Coding for wireless relay networks Conclusion
The Hamiltonian Quaternions: the definition
• Let {1,i,j,k}be a basis for a vector space of dim 4 over R. • We have the rule that i2 =−1,j2 =−1, andij =−ji. • The Hamiltonian Quaternionsis the set Hdefined by
H={x+yi +zj +wk |x,y,z,w ∈R}.
• Hamiltonian Quaternions are a division algebra:
q−1 = ¯q
qq¯,
Division algebras Coding for wireless relay networks Conclusion
The Hamiltonian Quaternions: the definition
• Let {1,i,j,k}be a basis for a vector space of dim 4 over R. • We have the rule that i2 =−1,j2 =−1, andij =−ji. • The Hamiltonian Quaternionsis the set Hdefined by
H={x+yi +zj +wk |x,y,z,w ∈R}.
• Hamiltonian Quaternions are a division algebra:
q−1 = ¯q
qq¯,
Division algebras Coding for wireless relay networks Conclusion
The Hamiltonian Quaternions: how to get matrices
• Any quaternion q=x+yi+zj+wk can be written as (x+yi) +j(z−wi) =α+jβ, α, β∈C.
• Now compute themultiplication byq: (α+jβ)
| {z }
q
(γ+jδ) = αγ+jαδ¯ +jβγ+j2βδ¯ = (αγ−βδ) +¯ j( ¯αδ+βγ)
• Write this equality in the basis {1,j}:
α −β¯ β α¯ γ δ = αγ−βδ¯ ¯ αδ+βγ
Division algebras Coding for wireless relay networks Conclusion
The Hamiltonian Quaternions: how to get matrices
• Any quaternion q=x+yi+zj+wk can be written as (x+yi) +j(z−wi) =α+jβ, α, β∈C.
• Now compute themultiplication byq: (α+jβ)
| {z }
q
(γ+jδ) = αγ+jαδ¯ +jβγ+j2βδ¯ = (αγ−βδ) +¯ j( ¯αδ+βγ)
• Write this equality in the basis {1,j}:
α −β¯ β α¯ γ δ = αγ−βδ¯ ¯ αδ+βγ
Division algebras Coding for wireless relay networks Conclusion
The Hamiltonian Quaternions: how to get matrices
• Any quaternion q=x+yi+zj+wk can be written as (x+yi) +j(z−wi) =α+jβ, α, β∈C.
• Now compute themultiplication byq: (α+jβ)
| {z }
q
(γ+jδ) = αγ+jαδ¯ +jβγ+j2βδ¯ = (αγ−βδ) +¯ j( ¯αδ+βγ)
• Write this equality in the basis {1,j}:
α −β¯ β α¯ γ δ = αγ−βδ¯ ¯ αδ+βγ
Division algebras Coding for wireless relay networks Conclusion
Introducing cyclic algebras
• The Hamiltonian Quaternions gives the Alamouti code:
q =α+jβ ∈H↔
α −β¯ β α¯
• We similarly consider cyclic algebras:
x=x0+ex1 ∈ A ↔
x0 γσ(x1)
x1 σ(x0)
Division algebras Coding for wireless relay networks Conclusion
Introducing cyclic algebras
• The Hamiltonian Quaternions gives the Alamouti code:
q =α+jβ ∈H↔
α −β¯ β α¯
• We similarly consider cyclic algebras:
x=x0+ex1 ∈ A ↔
x0 γσ(x1)
x1 σ(x0)
Division algebras Coding for wireless relay networks Conclusion
Advantages of cyclic algebras
1. Yield full diversityand a practical encoding (of n2 information symbols for ann×n codeword), for anynumber of antennas.
2. Allow for alower bound on the minimum determinant for constellations of arbitrary size.
3. Achieve the diversity-multiplexing tradeoff of Zheng and Tse, thanks to thenon-vanishing determinant property.
• Constructions of codes are available where the algebraic structures are exploited tooptimize the codes performance. F. E. Oggier, G. Rekaya, J.-C. Belfiore, E. Viterbo. Perfect Space-Time
Division algebras Coding for wireless relay networks Conclusion
Advantages of cyclic algebras
1. Yield full diversityand a practical encoding (of n2 information symbols for ann×n codeword), for anynumber of antennas. 2. Allow for alower bound on the minimum determinant for
constellations of arbitrary size.
3. Achieve the diversity-multiplexing tradeoff of Zheng and Tse, thanks to thenon-vanishing determinant property.
• Constructions of codes are available where the algebraic structures are exploited tooptimize the codes performance. F. E. Oggier, G. Rekaya, J.-C. Belfiore, E. Viterbo. Perfect Space-Time
Division algebras Coding for wireless relay networks Conclusion
Advantages of cyclic algebras
1. Yield full diversityand a practical encoding (of n2 information symbols for ann×n codeword), for anynumber of antennas. 2. Allow for alower bound on the minimum determinant for
constellations of arbitrary size.
3. Achieve the diversity-multiplexing tradeoff of Zheng and Tse, thanks to thenon-vanishing determinant property.
• Constructions of codes are available where the algebraic structures are exploited tooptimize the codes performance. F. E. Oggier, G. Rekaya, J.-C. Belfiore, E. Viterbo. Perfect Space-Time
Division algebras Coding for wireless relay networks Conclusion
Advantages of cyclic algebras
1. Yield full diversityand a practical encoding (of n2 information symbols for ann×n codeword), for anynumber of antennas. 2. Allow for alower bound on the minimum determinant for
constellations of arbitrary size.
3. Achieve the diversity-multiplexing tradeoff of Zheng and Tse, thanks to thenon-vanishing determinant property.
• Constructions of codes are available where the algebraic structures are exploited tooptimize the codes performance. F. E. Oggier, G. Rekaya, J.-C. Belfiore, E. Viterbo. Perfect Space-Time
Division algebras Coding for wireless relay networks Conclusion
Performances
• A 2×2 cyclic algebra based code is to be implemented in the future wireless standard 802.16e for wireless LANs.
Division algebras Coding for wireless relay networks Conclusion
Non-coherent unitary space-time coding
• We assume no channel knowledge.
• Use a cyclic division algebra endowed with an involution:
A Mn(L)
x ↔ X
α(x) ↔ X†
xα(x) = 1 ↔ XX†=I
F. Oggier,Cyclic Algebras for Noncoherent Differential Space-Time Coding. Trans. on IT.
• Use the Cayley transformof an Hermitian matrix A:
X= (I+iA)−1(I−iA).
F. Oggier, B. Hassibi,Algebraic Cayley differential Space-Time Codes. Trans. on IT.
Division algebras Coding for wireless relay networks Conclusion
Non-coherent unitary space-time coding
• We assume no channel knowledge.
• Use a cyclic division algebra endowed with an involution:
A Mn(L)
x ↔ X
α(x) ↔ X†
xα(x) = 1 ↔ XX†=I
F. Oggier,Cyclic Algebras for Noncoherent Differential Space-Time Coding. Trans. on IT.
• Use the Cayley transformof an Hermitian matrix A:
X= (I+iA)−1(I−iA).
F. Oggier, B. Hassibi,Algebraic Cayley differential Space-Time Codes. Trans. on IT.
Division algebras Coding for wireless relay networks Conclusion
Non-coherent unitary space-time coding
• We assume no channel knowledge.
• Use a cyclic division algebra endowed with an involution:
A Mn(L)
x ↔ X
α(x) ↔ X†
xα(x) = 1 ↔ XX†=I
F. Oggier,Cyclic Algebras for Noncoherent Differential Space-Time Coding. Trans. on IT.
• Use the Cayley transformof an Hermitian matrix A:
X= (I+iA)−1(I−iA).
F. Oggier, B. Hassibi,Algebraic Cayley differential Space-Time Codes. Trans. on IT.
Division algebras Coding for wireless relay networks Conclusion
Performances
10 12 14 16 18 20 22 24 26 28 30 10−1 algebraic orthogonal design cayley code 1Division algebras Coding for wireless relay networks Conclusion
Division algebras
Motivation: coherent space-time coding Introducing division algebras
Other space-time coding applications
Coding for wireless relay networks Distributed space-time coding
Noncoherent distributed space-time coding
Conclusion Future work
Division algebras Coding for wireless relay networks Conclusion
Some references
1. J.N. Laneman and G. W. Wornell, “Distributed
space-time-coded protocols for exploiting cooperative diversity in wireless network”, Oct. 2003.
2. K. Azarian, H. El Gamal and P. Schniter,“On the achievable diversity-multiplexing tradeoff in half-duplex cooperative channels”, Dec. 2005.
3. Y. Jing and B. Hassibi, “Distributed space-time coding in Wireless Relay Networks”, Dec. 2006.
4. S. Yang and J.-C. Belfiore, “ Optimal space-time codes for the MIMO Amplify-and-Forward cooperative channel”, Feb. 2007.
Division algebras Coding for wireless relay networks Conclusion
Coding for wireless relay network
• Relay nodes are small devices with few resources.
• Cooperation: we learntdiversityfrom space-time coding.
Tx
Rx
Division algebras Coding for wireless relay networks Conclusion
Wireless relay network: phase 1
• s vector for one Tx antenna, matrix for several Tx antennas.
Tx Rx s r1=f1s+v1 r2=f2s+v2 r3=f3s+v3 r4=f4s+v4 1
Division algebras Coding for wireless relay networks Conclusion
Wireless relay network: phase 2
• No decoding at the relays. • The matrices Ai are unitary.
Tx Rx s A1r1 A2r2 A3r3 A4r4 t1 t2 t3 t4 1
Division algebras Coding for wireless relay networks Conclusion
Channel model
1. At the receiver, y= R X i=1 giti+w= R X i=1 giAi(sfi+vi) +w 2. So that finally: y= [A1s· · ·ARs] | {z } X f1g1 .. . fngn | {z } H +w0• Each relay encodes a (set of) column(s), so that the encoding is distributedamong the nodes (Jing-Hassibi).
Division algebras Coding for wireless relay networks Conclusion
Channel model
1. At the receiver, y= R X i=1 giti+w= R X i=1 giAi(sfi+vi) +w 2. So that finally: y= [A1s· · ·ARs] | {z } X f1g1 .. . fngn | {z } H +w0• Each relay encodes a (set of) column(s), so that the encoding is distributedamong the nodes (Jing-Hassibi).
Division algebras Coding for wireless relay networks Conclusion
Channel model
1. At the receiver, y= R X i=1 giti+w= R X i=1 giAi(sfi+vi) +w 2. So that finally: y= [A1s· · ·ARs] | {z } X f1g1 .. . fngn | {z } H +w0• Each relay encodes a (set of) column(s), so that the encoding is distributedamong the nodes (Jing-Hassibi).
Division algebras Coding for wireless relay networks Conclusion
Distributed space-time codes from division algebras
• σ :L→L,σ4 = 1, 4·4 information symbols, doesnotwork!
[A1s A2s A3s A4s] [A1S A2S] 6 = x0 iσ(x3) iσ2(x2) iσ3(x1) x1 σ(x0) iσ2(x3) iσ3(x2) x2 σ(x1) σ2(x0) iσ3(x3) x3 σ(x2) σ2(x1) σ3(x0) • τ :K →K,K ⊂L,τ2 = 1 xi=information symbols x0 iτ(x3) ix2 iτ(x1) x1 τ(x0) ix3 iτ(x2) x2 τ(x1) x0 iτ(x3) x3 τ(x2) x1 τ(x0) , x0 ix3 ix2 ix1 x1 x0 ix3 ix2 x2 x1 x0 ix3 x3 x2 x1 x0
F. Oggier, B. Hassibi,An Algebraic Coding Scheme for Wireless Relay Networks with Multiple-Antenna Nodes
Division algebras Coding for wireless relay networks Conclusion
Distributed space-time codes from division algebras
• σ :L→L,σ4 = 1, 4·4 information symbols, doesnotwork!
[A1s A2s A3s A4s] [A1S A2S] 6 = x0 iσ(x3) iσ2(x2) iσ3(x1) x1 σ(x0) iσ2(x3) iσ3(x2) x2 σ(x1) σ2(x0) iσ3(x3) x3 σ(x2) σ2(x1) σ3(x0) • τ :K →K,K ⊂L,τ2 = 1 xi=information symbols x0 iτ(x3) ix2 iτ(x1) x1 τ(x0) ix3 iτ(x2) x2 τ(x1) x0 iτ(x3) x3 τ(x2) x1 τ(x0) , x0 ix3 ix2 ix1 x1 x0 ix3 ix2 x2 x1 x0 ix3 x3 x2 x1 x0
F. Oggier, B. Hassibi,An Algebraic Coding Scheme for Wireless Relay Networks with Multiple-Antenna Nodes
Division algebras Coding for wireless relay networks Conclusion
Performances
16 18 20 22 24 26 28 30 10−4 10−3 10−2 10−1 100 P(dB) BLER M=N=R=2 algebraic random no coding 1Division algebras Coding for wireless relay networks Conclusion
Division algebras
Motivation: coherent space-time coding Introducing division algebras
Other space-time coding applications
Coding for wireless relay networks Distributed space-time coding
Noncoherent distributed space-time coding
Conclusion Future work
Division algebras Coding for wireless relay networks Conclusion
A Noncoherent Channel
• How to design a protocol to communicate over a wireless relay network with no channel information?
• Noncoherent network model: y= [A1s· · ·ARs]
| {z } Xunitary f1g1 .. . fngn | {z } H +w0 • Design s0= √1 T(1, . . . ,1) t, s i =Uis0, i = 1, . . . ,L,where Ui’s areT ×T unitary matrices.
Division algebras Coding for wireless relay networks Conclusion
A Noncoherent Channel
• How to design a protocol to communicate over a wireless relay network with no channel information?
• Noncoherent network model: y= [A1s· · ·ARs]
| {z } Xunitary f1g1 .. . fngn | {z } H +w0 • Design s0= √1 T(1, . . . ,1) t, s i =Uis0, i = 1, . . . ,L,where Ui’s areT ×T unitary matrices.
Division algebras Coding for wireless relay networks Conclusion
A Noncoherent Channel
• How to design a protocol to communicate over a wireless relay network with no channel information?
• Noncoherent network model: y= [A1s· · ·ARs]
| {z } Xunitary f1g1 .. . fngn | {z } H +w0 • Design s0= √1 T(1, . . . ,1) t, s i =Uis0, i = 1, . . . ,L,where Ui’s areT ×T unitary matrices.
Division algebras Coding for wireless relay networks Conclusion
A Unitary Distributed Space-Time Code
1. Let M be aT ×T matrixM such that MM†=T. Then
Ai =diag( Mi |{z} ith column
), i = 1, . . . ,R.
2. Choosing allUj diagonal (tocommute with allAi)
[A1sj, . . . ,ARsj] = [UjA1s0, . . . ,UjARs0] =UjM/
√ T,
Division algebras Coding for wireless relay networks Conclusion
A Unitary Distributed Space-Time Code
1. Let M be aT ×T matrixM such that MM†=T. Then
Ai =diag( Mi |{z} ith column
), i = 1, . . . ,R.
2. Choosing allUj diagonal (tocommute with allAi)
[A1sj, . . . ,ARsj] = [UjA1s0, . . . ,UjARs0] =UjM/
√ T,
Division algebras Coding for wireless relay networks Conclusion
A Unitary Distributed Space-Time Code
1. Let M be aT ×T matrixM such that MM†=T. Then
Ai =diag( Mi |{z} ith column
), i = 1, . . . ,R.
2. Choosing allUj diagonal (tocommute with allAi)
[A1sj, . . . ,ARsj] = [UjA1s0, . . . ,UjARs0] =UjM/
√ T,
Division algebras Coding for wireless relay networks Conclusion
Butson-Hadamard Matrices
1. A Generalized Butson Hadamard(GBH) matrix is a T ×T
matrix M such that
MM∗ =M∗M =TIT
wheremij∗ =m−ji1.
2. Choose the coefficients of M to beroots of unity. Furthermore all Ai are unitary.
3. Let ζ3 = exp(2iπ/3) be a primitive 3rd root of unity.
M = 1 1 1 1 ζ3 ζ32 1 ζ32 ζ3
Division algebras Coding for wireless relay networks Conclusion
Butson-Hadamard Matrices
1. A Generalized Butson Hadamard(GBH) matrix is a T ×T
matrix M such that
MM∗ =M∗M =TIT
wheremij∗ =m−ji1.
2. Choose the coefficients of M to beroots of unity. Furthermore all Ai are unitary.
3. Let ζ3 = exp(2iπ/3) be a primitive 3rd root of unity.
M = 1 1 1 1 ζ3 ζ32 1 ζ32 ζ3
Division algebras Coding for wireless relay networks Conclusion
Butson-Hadamard Matrices
1. A Generalized Butson Hadamard(GBH) matrix is a T ×T
matrix M such that
MM∗ =M∗M =TIT
wheremij∗ =m−ji1.
2. Choose the coefficients of M to beroots of unity. Furthermore all Ai are unitary.
3. Let ζ3 = exp(2iπ/3) be a primitive 3rd root of unity.
M = 1 1 1 1 ζ3 ζ32 1 ζ32 ζ3
Division algebras Coding for wireless relay networks Conclusion
A Differential Encoder
1. Transmitter: send st =s0 then st+T =U(zt+T)
| {z } data st. 2. Relays: ri(t) =fist+vi(t), ri(t+T) =fiU(zt+T)st+vi(t+T). 3. Receiver: y(t) =PR i=1gifiAist+w(t). y(t+T) =PR i=1gifiAiU(zt+T)st+w(t+T).
4. The differential channel:
Division algebras Coding for wireless relay networks Conclusion
A Differential Encoder
1. Transmitter: send st =s0 then st+T =U(zt+T)
| {z } data st. 2. Relays: ri(t) =fist+vi(t), ri(t+T) =fiU(zt+T)st+vi(t+T). 3. Receiver: y(t) =PR i=1gifiAist+w(t). y(t+T) =PR i=1gifiAiU(zt+T)st+w(t+T).
4. The differential channel:
Division algebras Coding for wireless relay networks Conclusion
A PEP computation
1. We consider a mismatched decoder.
2. We prove that
P(Uk →Ul)≤Egdet(IT +8(c0 cρ
ρkgk2+1)D|g|(Uk −Ul)(Uk −Ul) †)−1.
3. So that, using a result by Jing-Hassibi, the diversity is
R 1−log logP logP , when (Uk−Ul)(Uk−Ul)† is full rank.
Division algebras Coding for wireless relay networks Conclusion
A PEP computation
1. We consider a mismatched decoder. 2. We prove that
P(Uk →Ul)≤Egdet(IT +8(c0 cρ
ρkgk2+1)D|g|(Uk −Ul)(Uk −Ul) †)−1.
3. So that, using a result by Jing-Hassibi, the diversity is
R 1−log logP logP , when (Uk−Ul)(Uk−Ul)† is full rank.
Division algebras Coding for wireless relay networks Conclusion
A PEP computation
1. We consider a mismatched decoder. 2. We prove that
P(Uk →Ul)≤Egdet(IT +8(c0 cρ
ρkgk2+1)D|g|(Uk −Ul)(Uk −Ul) †)−1.
3. So that, using a result by Jing-Hassibi, the diversity is
R 1−log logP logP , when (Uk−Ul)(Uk−Ul)† is full rank.
Division algebras Coding for wireless relay networks Conclusion
Code Design
1. We want to design unitary diagonalmatricesU,independent of the matrices at the relays.
2. They have to satisfy
det(U(zt)−U(zt0))6= 0, t 6=t0.
3. Usecyclic codes(Hochwald and Sweldens):
ζu1l L 0 0 . .. 0 0 ζuMl L , l = 0, . . . ,L−1, ζL= exp(2iπ/L)
whereL andu = (u1, . . . ,uM) have to be designed.
F. Oggier, B. Hassibi,A Coding Strategy for Wireless Networks with no Channel Information
Division algebras Coding for wireless relay networks Conclusion
Code Design
1. We want to design unitary diagonalmatricesU,independent of the matrices at the relays.
2. They have to satisfy
det(U(zt)−U(zt0))6= 0, t 6=t0.
3. Usecyclic codes(Hochwald and Sweldens):
ζu1l L 0 0 . .. 0 0 ζuMl L , l = 0, . . . ,L−1, ζL= exp(2iπ/L)
whereL andu = (u1, . . . ,uM) have to be designed.
F. Oggier, B. Hassibi,A Coding Strategy for Wireless Networks with no Channel Information
Division algebras Coding for wireless relay networks Conclusion
Performances
12 14 16 18 20 22 24 26 10−8 10−7 10−6 10−5 10−4 10−3 10−2 10−1 100 P(dB) BLER R=3 R=6 R=9Division algebras Coding for wireless relay networks Conclusion
Conclusion...
• The problem of designing fully-diversematrices arise in a lot of wireless coding applications, and division algebras is a powerful tool to design such matrices.
• We thus could propose codes forspace-time coding as well as wireless relay networks.
• We also propose a strategy for communicate over a wireless relay network with no channel information.
Division algebras Coding for wireless relay networks Conclusion
Conclusion...
• The problem of designing fully-diversematrices arise in a lot of wireless coding applications, and division algebras is a powerful tool to design such matrices.
• We thus could propose codes forspace-time coding as well as wireless relay networks.
• We also propose a strategy for communicate over a wireless relay network with no channel information.
Division algebras Coding for wireless relay networks Conclusion
Conclusion...
• The problem of designing fully-diversematrices arise in a lot of wireless coding applications, and division algebras is a powerful tool to design such matrices.
• We thus could propose codes forspace-time coding as well as wireless relay networks.
• We also propose a strategy for communicate over a wireless relay network with no channel information.
Division algebras Coding for wireless relay networks Conclusion
...and future work
• Can we do better for non-coherent wireless networks?
• Synchronization is an issue to be dealt with.
• Future work is also oriented towards including securityin wireless communication.
Division algebras Coding for wireless relay networks Conclusion
...and future work
• Can we do better for non-coherent wireless networks? • Synchronization is an issue to be dealt with.
• Future work is also oriented towards including securityin wireless communication.
Division algebras Coding for wireless relay networks Conclusion
...and future work
• Can we do better for non-coherent wireless networks? • Synchronization is an issue to be dealt with.
• Future work is also oriented towards including securityin wireless communication.
Division algebras Coding for wireless relay networks Conclusion