Академический Документы
Профессиональный Документы
Культура Документы
and further wireless systems where the channel is mostly s(t) = am,n gm,n (t), (1)
doubly dispersive. Contrary to the classic OFDM system n=−∞ m=0
2
this paper via the efficient implementation method derived
1
in [7]. ξ=1
Heisenberg parameter ξ
written as
Direction parameter η
r(t) = h(t, τ )s(t − τ )dτ = H(ν, τ )s(t − τ )ej2πνt dνdτ
ξ (λ=1)
1
then for OFDM/OQAM (5) can be rewritten as (8), shown frequency plane. The larger ξ is, the smaller space the pulse
on the top of next page. Under the assumption of wide shape occupies in the T-F plane.
sense stationary uncorrelated scattering (WSSUS) channel,
we have (Δt)2 = R t2 |g(t)|2 dt
(9)
(Δf )2 = R f 2 |G(f )|2 df
E{H(ν, τ )H ∗ (ν , τ )} = Sh (τ, ν)δ(τ − τ , ν − ν )
(6)
Here g(t) is assumed to be origin-centered with unity en-
˙
where E{} is the expectation operator and Sh (τ, ν) is the ergy [8] for simple expressions.
channel scattering function. Assume all the transmitted
∗
symbols are independent with uniform energy, i.e., E{am,n am ,n } =
In order to know how the pulse shape spreads over the T-
δmm δnn , and apply the WSSUS assumption (6), the energy F plane, we define the Direction parameter η = Δf Δt
. For
of the desired signal part S and the interference part I in
τ0 1 ν0 λ
(8) can be written as EGF functions with T = 2λ and F = 2 , where λ > 0
∗
2
is a constant scaling factor, the variation of ξ and η with
ES = E{SS
} = Sh (τ, ν)|Ag (τ, ν)| dνdτ respect to α for EGF functions with λ = 1 ( τT0 = νF0 =
2 (7)
√
EI = (m,n)=(0,0) Sh (τ, ν)|Ag (nτ0 + τ, mν0 + ν)| dνdτ 2
2
) and λ = 2 ( τT0 = 12 , νF0 = 1) is shown in Fig. 1, in
which are the same as, at least on the analogy of, the energy which τ0 and ν0 are normalised by T and F respectively √for
expressions derived for pulse shape multicarrier systems [10] convenience. Compared to the case with τT0 = νF0 = 22 ,
and for hexagonal multicarrier systems [9]. Different optimi- the EGF function with τT0 = 12 have larger variation of η
sation methods regarding maximising desired signal energy and better stability of ξ, which makes it more suitable for
ES [11], or minimising interference EI [10, 9], or maximising pulse shape adaptation. The Gaussian function gα (t) which
the signal to interference ratio ES /EI [10] are considered. has η = 1, will not be taken into consideration since it will
However analytical solutions only exist for some special cases introduce large reconstruction distortion as we will see later.
and therefore numerical solutions are used for general cases.
3. TIME FREQUENCY LOCALISATION PA- To maximise the immunity to delay and frequency disper-
sion, the optimum pulse shape should have the same shape
RAMETERS as the channel itself [2, 11], namely,
For different channels, the optimal pulse shape is normally
different. A widely used parameter to measure the time Δt τrms T
frequency localization of the pulse shape is the Heisenberg ≈ ≈ (10)
1
Δf fD F
parameter [2] ξ = 4πΔtΔf ≤ 1 with its maximum achieved
2 where τrms is the root-mean-square (RMS) delay spread and
by the Gaussian function gα (t) = (2α)1/4 e−παt , α > 0. Δt
is the mass moment of inertia of the prototype function in fD is the maximum Doppler shift. If the value of η is calcu-
time and Δf in frequency, which indicates how the energy lated with normalised τ0 and ν0 (by T and F respectively)
(mass) of the prototype function spreads over the time and as in Fig. 1, (10) can be rewritten as
1 τrms /T τrms Fs 2
There is another definition for the ambiguity function,
which differs by a phase shift. η(α) ≈ = ( ) (11)
fD /F fD N
â0,0 = am,n j
m,n
m+n
H(ν, τ )Ag (nτ0 + τ, mν0 + ν)ejπ(nτ0 +τ)(mν
0 +ν)
dνdτ
= a0,0 H(ν, τ )Ag (τ, ν)ejπτ ν
dνdτ + am,n j m+n
H(ν, τ )Ag (nτ0 + τ, mν0 + ν)ejπ(nτ0 +τ )(mν0 +ν) dνdτ
(8)
(m,n)=(0,0)
Signal S
Interference I
5
Therefore, for each specific channel realisation (i.e. τfrms
D
is
determined), the performance against delay and Doppler dis- 0
persion depends on the bandwidth F and the direction pa-
rameter η. We can adjust these two parameters to improve −5
the system performance. When the sampling frequency Fs
is fixed in some instance, the FFT size N will be subject to −10
0
spectrum, we have 10
EGF 0.5
|τ | EGF Es
−τ EGF Ei
e rms 1
Sh (τ, ν) = (15) OFDM 1/8
halfcsine
τrms πfD 1 − ( fνD )2
Uncoded BER
Table 2. For a carrier frequency fc = 2.5GHz, Doppler
spread Bd = 2fD = 700Hz is equivalent to a moving speed
of 157.5km/h.
−2
10
[3] P. Jung, G. Wunder and C. S. Wang, “OQAM/IOTA
Downlink Air Interface for UMTS HSDPA Evolution,”
9th International OFDM-Workshop, Hamburg, pp.
153–157, 2004.
0
10
EGF 0.5
[4] M. Bellec and P. Pirat, “OQAM performances and
EGF Es complexity,” IEEE P802.22 Wireless Regional Area
EGF Ei
OFDM 1/8 Network, Jan. 2006.
halfcsine
[5] TIA Committee TR-8.5, “Wideband Air Interface
Isotropic Orthogonal Transform Algorithm (IOTA)
−1
10
–Public Safety Wideband Data Standards Project –
Digital Radio Technical Standards,” TIA-902.BBAB
(Physical Layer Specification, Mar. 2003) and
Uncoded BER