System for supporting low-latency extended reality services over ethernet passive optical network and a method thereof
The present invention discloses a network system for supporting XR devices/users over an EPON comprising one or more user XR devices, an edge server and a EPON based connection between said XR devices and said edge server to offer XR services with strict latency bounds over said EPON based connection involving offloading of XR device generated data to the edge server for processing, whereby the edge server process the XR data based on predicted future XR frames of said data after reconstructing EPON dependent inter-arrival pattern of the XR frames.
1 . A network system for supporting XR devices/users over an EPON comprising
one or more user XR devices;
an edge server;
EPON based connection between said XR devices and said edge server to offer XR services with strict latency bounds over said EPON based connection involving offloading of XR device generated data to the edge server for processing, whereby the edge server processes the XR data based on predicted future XR frames of said data after reconstructing inter-arrival pattern of the XR frames;
wherein the edge server is updated with queuing delay of each XR frame due to the EPON by involving periodic generation time of each XR frame at the XR device to make the reconstruction of the inter-arrival pattern feasible;
wherein the edge server includes a play-off buffer which stores uplink XR data to add an extra play-off delay to each XR frame (denoted by
D
b
k
,
x
for x th frame of kth XK user), making the total queuing (denoted by
D
q
k
,
x
)
and play-off delay a constant
(
say
D
po
k
=
D
q
k
+
D
b
k
for
k
th
XR
user
)
;
wherein total uplink delay faced by all the XR frames of the k th user XR device
(
D
tot
k
)
is given by:
D
tot
k
=
D
prop
k
+
D
tx
k
+
D
po
k
≤
D
ub
k
+
D
p
k
where
D
prop
k
,
D
tx
k
,
and
D
ub
k
denote propagation time, transmission time, and uplink delay bound respectively, here improved inter-arrival pattern reconstruction is achievable with greater
D
po
k
,
which lowers prediction error and to increase
D
po
k
,
D
p
k
is needed to be increased as well, resulting in an increment of prediction error; and
wherein the edge server involves the play-off buffer to set best
D
p
k
to reduce the prediction error.
2 . The network system as claimed in claim 1 , wherein the edge server includes
a data receiver for receiving the XR data;
said play-off buffer to store the XR data for reconstruction of the inter-arrival pattern of the XR frames as generated by the XR device user;
a reconstruction controller for controlling operation of the play-off buffer to play the stored frames at a selected speed ensuring all the frames are subject to a constant delay;
an AI based frame predictor to predict future frames based on the reconstructed data;
a processor for processing thus predicted future frames; and
a transmitter for transmitting the processed XR frames to the one or more XR device users.
3 . The network system as claimed in claim 1 , wherein the EPON based connection includes OLU and OLT based connection.
4 . The network system as claimed in claim 3 , includes SDN controller operating in combination with the edge server, the OLT and the ONU for facilitating the edge server to reselect the prediction time for all XR device users every time when an XR device users registers or deregisters which changes the network load and also changing error in the reconstruction pattern.
5 . The network system as claimed in claim 1 , wherein the edge server further includes prediction duration calculator for selecting prediction duration for the all XR devices and informing the same to the SDN controller and the AI-based frame predictor.
6 . The network system as claimed in claim 4 , wherein the SDN controller includes class mapper which decides class corresponding to delay bound of
D
u
b
k
+
D
p
k
,
and informs the class information to the OLT and ONUs enabling the OLT schedules the XR device users.
7 . The network system as claimed in claim 1 , wherein the playoff-buffers are implemented using edge server memory and the reconstruction controller, AI-based frame predictor and the Prediction duration calculator are implemented involving processing, memory, and storage of the edge server.
8 . A method for XR devices/users over an EPON comprising
sending registration request to an edge server by an XR device user including informing its requirement to a SDN controller;
allocating a default class (correspond to
D
p
k
=
0
)
according to requirement by class mapper of the SDN controller and informing the same to OLTs and ONUs;
initiating uplink transmission of XR data by the XR device though the OLTs and ONUs which are received by receiver of the edge server;
storing the XR data in play-off buffer of the edge server to reconstruct inter-arrival pattern of XR frames of data as generated by user of the XR device;
controlling the play-off buffer by a reconstruction controller to play the XR frames ensuring all frames are subject to a constant delay;
predicting future frames based on the reconstructed data using an AI-based frame predictor;
processing the predicted frames at a processor and after processing, sending the processed data to the XR device users by a transmitter;
involving a prediction duration calculator for selecting prediction durations and informing the same to the SDN controller and the AI-based frame predictor; and
involving a class mapper in the SDN controller to decide class corresponding to delay bound of
D
u
b
k
+
D
p
k
,
and informs the class information to the OLT and the ONUs for scheduling the XR device users.