To address the difficulty to reconcile latency, bandwidth/utilization, and stability, we believe a good design of CC is the key. Unfortunately, we found state-of-art CC mechanisms in RDMA networks, such as DCQCN [43] and TIMELY [31], have some essential limitations:
Despite these potentials, commercial 5G services are at their infancy.
In contrast, different entities involved in the streaming system such as the content provider, ISP and app developers have traditionally possessed only partial views and optimized specific factors somewhat independently, based mainly on their limited views.
Additionally
On the flip side / On the other hand, …
In addition, HPCC also has the following essential properties for being practical:
The partitions in the log serve several purposes. First, …
Interestingly, we also find that the (in)activity of user traffic can trigger 4G-5G handoffs.
Other than that, being an agent for all the on-board passengers, a mobile relay should optimize the aggregated user experience rather than any single user, while there is currently no practical solution towards this goal, especially from an application-agnostic perspective.
It is worth mentioning that compared to HTTPS, QUIC has some unique properties which make its analysis even more challenging
Because
fix itself cannot be defined in simply typed lambda-calculus. Indeed, no expression that can lead to non-termination computations can be typed using only simple type.
Due to the discrepancy between 4G and 5G, frequently switching between them may confuse applications and bring highly inconsistent user experiences.
We attribute this to the clear LoS and sufficiently high transmis- sion power of 5G antennas.
This is likely because NSA uses 4G for control-plane signaling
This is attributed to two reasons.
At a high level, the aforementioned problems arise as a result of overlaying the bit-rate adaptation algorithms on top of several logical layers of the network stack.
Comparison
Unlike existing approaches that often require a large number of iterations to find the proper flow rates, HPCC requires only one rate update step in most cases.
Compared to 4G LTE, 5G offers significantly higher bandwidth, lower latency, and better scalability (i.e., supporting more devices).
NSA is contrasted with a Standalone deployment (SA), which is fully independent of legacy cellular infrastructures.
Due to the ultra-high bandwidth of 5G, the bottleneck of an end-to-end path may easily shift from the wireless hop to the Internet – a situation that seldom appears in 3G/4G.
The former provides multi-Gbps rates but is susceptible to blockage, whereas the latter is slower but offers reliable connectivity.
Solution
To overcome this, mmWave transceivers have to use phased-array antennas to form highly directional beams.
The principles for recovery are 1) reducing the number of interruptions for QoS satisfaction and 2) reducing the number of recoveries for avoiding overheads.
Consequence
We also experimentally show that 5G’s high throughput does not always translate to better app QoE, whose improvement requires joint, cross-layer optimizations from multiple players in the mobile ecosystem. We make the following contributions in this paper.
As shown, both obstructions trigger 5G-to-4G handoffs and lead to significant performance degradation.